You might want to read through this:
https://en.bitcoin.it/wiki/Difficulty#What_is_the_maximum_difficulty.3F

On a more technical level - those miner developers that check for a fixed value and find that at the speed they're hashing they're finding enough valid nonces that the host system is getting taxed (as the host checks whether the used nonce results in 1. a hash that counts as a valid share for a given pool and/or 2. a hash that meets the Bitcoin block target difficulty, will probably adjust their hardware to check for a more difficult fixed value.  In essence something that's already been done at pools for their share difficulty, as very high hash rate machines were just hammering the network connections.  We're a long, long way from even that being an issue on a Bitcoin block target though.
The leading zeros description is a simplification, but even just taking that, try working out the effect on the difficulty of adding a leading zero requirement, and what that would mean for the average network hash rate in order to (ideally) find a solution every 10 minutes.

The situation you 'predicted' is entirely academic.

The only reason the target would be so low is because the overall network speed is high enough that blocks get solved on average every 10 minutes the last 2016 blocks.  The only reason it would 'not be found' would be because the overall network speed would drop dramatically.

Can you think of any practical situation in which that would occur?

Or it might go up by a few percent: https://bitcoinwisdom.com/bitcoin/difficulty

The situation sketched by the OP though was a theoretical one, and one that in reality wouldn't occur - an unreachable difficulty would imply an unreachable hash rate existed before that.  Even if it did occur - i.e. a whole period without a single block mined, say because every hardware miner suddenly disappeared from existence - the next period would have difficulty reset to scratch, and you'd finally be able to put this to practical use: http://gizmodo.com/mining-bitcoin-with-pencil-and-paper-1640353309

I have no idea where you're getting that notion.  On a larger overview of history, power consumption went down for a while as ASICs were introduced that were far more efficient giving vastly more hashing power for a given power consumption, which drowned out all the power-hungry CPUs and GPUs, going back up again because more of those same ASICs were being throw into mining, back down again a bit because they got more efficient, back up because more miners but not as much because a lot of people are trying to mine more for power efficiency than maximum hashing power.
Also, this is completely unrelated to what we were talking about, so let's move on...

And this is where you'll have to explain the mathematical reason that you're thinking of that would lead to that situation.

So what I guess you're asking is: should the impossible (or, depending on your mathematical reason: highly unlikely but academically possible) occur, is there a fail safe for difficulty adjustment that is bound to something other than the block chain length?  In which case, I believe the answer is that there currently is no such fail safe, but I'd suspect one would get added if there were reason to believe it were imminent - and probably no sooner.

There are existing discussions on this elsewhere, just google for something like 'bitcoin unsolvable block' or 'minimum target' / 'maximum difficulty', though none deal with how to deal with that issue ... simply because every discussion concedes that it's impossible to ever get into that state ('blind zone' per your terms); something you'll have to counter

Note that one situation which is theorized to exist is that for a given block header, there is no nonce that will yield a hash that falls below the target difficulty.  But that's not an issue, because there's other fields that can be manipulated, transactions can be moved around, excluded, included, etc.  One practical example of this lies in ASIC chip design.  I don't remember whose it was, but one of them has a known limited nonce range making it more likely to, eventually, be unable to solve a block without adjusting inputs.  Of course by then that hardware is obsolete anyway.

Especially when the halving occurs. I see miners turning off their units, and possibly a rise in the btc price to compensate for the halving.

This will happen when the zombie apocalypse comes.  But then, no one will give a flip about BitCoin.  Ill be up on the roof, scoring spectacular headshots with my Ruger 10/22 .22 rifle.

That still wouldn't cause the situation suggested by newIndia.

Simplisticly speaking: let's say all the farms, pools and individual miners (~330Phash/s) shut down... to the point where a single Block Erupter USB (~330Mhash/s or 1/1,000,000,000th of the hash rate) gets on average 10 minutes per block, and the difficulty adjusts to exactly that level.
Then all of those farms and pools come back online, and all 2016 blocks for that period end up getting mined not in 2 weeks (20,160 minutes), but in 20,160 minutes/1,000,000,0000 ~= 0.0012 second. (there's very practical reasons why that, currently, can't even happen, but let's roll with it).
The difficulty adjusts for that by making the difficulty.. 1,000,000,000 times as high - and we're back to 10 minutes per block for ~330Phash/s.

The difficulty adjustment is purely based on the previous period and a target; It doesn't attempt to compensate for any relative growth/decline.

A coin that did do so in a very naive way would have a bad time in the above scenario, as it could interpret the 1,000,000,000 times as high thing as a trend (expecting it to be another 1,000,000,000 times as high for the next period) and compensate by setting the difficulty 1,000,000,000,000,000,000 times as high.  But..that doesn't apply to Bitcoin.

Sure, but such a scenario is 1: highly unlikely (and OP never bothered to clarify their thoughts on this) and 2: just because it would take a long time, doesn't mean it wouldn't still occur.  That's not the same as 'unreachable', not even by a long shot.

---

Continuation from other thread:
If, hypothetical, if - all of which highly unlikely at best.

Paint me a picture of situations in which the above may occur in the Bitcoin network, assign a probability factor.

Even in one such highly unlikely situation, you said it yourself:
Emphasis mine - that's not the same as "not enough hashrate to solve new blocks"

I'm not sure pointing to altcoins - the majority of which that have failed have been mine-it-for-all-its-worth-then-dump-it - makes for a strong anecdotal case, but in terms of Bitcoin your argument seems to lean mostly on the reward halving causing a mass exodus of miners to the point where block intervals are sufficiently long that additional miners also lose interest.  Fair enough.  I still don't think that makes it unreachable/impossible to mine/etc; OP's original argument was that things would effectively be permanently stuck.  Whether or not it kills off a coin as a viable [insert marketing term du jour] is another matter.  We'll find out the answer to that soon enough

Remember that Bitcoin difficulty can only change by up to a factor of 4 with each difficulty reset.  That means that if the difficulty were 1, it could only increase to 4 on the next reset.  Then 16, then 64, then 256, etc.

Same thing coming down.  If the difficulty were 10 billion, it could only drop to as low as 2.5 billion on one reset, then to 625 million, and so on.