Are a blockchain reorganization attack and a 51% attack the same thing?
Slightly different. The former refer to type of the attack, while the latter refer to how to perform an attack.
Is 51% of the hashing power necessary for a reorg attack?
No. But 51% of hashrate ensure success of your attack.
An attacker is not required to have 51% in order to force a reorg, but the probability of success drops very quickly with lower hash rate and more blocks. This is why it is recommended to wait for additional confirmations if there is a risk due to a reorg.
They could force a reorg if they own most of the networking infrastructure, since Bitcoin Core packets are unencrypted (unless you use Tor to connect to it), they will be able to read any packet they want and drop some of them, preventing the relaying to a large portion of nodes of mined blocks that the operator don't like until other blocks are mined.
Jam enough blocks, and you can cause a mempool congestion, possibly even dropping transactions if the total size is over 300MB.
Even Hetzner could (theoretically) do this as they own a large percentage of nodes. If it fails, it would cause a chain split and would be even more destructive than a few reorged blocks (as transactions inside would go back into the mempool).
I'm not even sure Hetzner could theoretically do that when,
1. Most pool have it's own infrastructure, especially for fast block propagation (e.g. FIBRE and Falcon).
2. Decent amount of node (which accept incoming connection) use Tor.
3. Only 6.1% (which accept incoming connection) Hetzner.
4. Bitcoin Core connect to 10 other nodes and try to diversify it based on IP block.
P.S. I use
https://bitnodes.io/dashboard/ as primary reference.