There are a couple of real leads beyond FLOSSmole.
First, Bryan Bishop (kanzure) is hosting a big tarball of IRC logs here:
https://diyhpl.us/~bryan/irc/bitcoin/bitcoin-logs.2010-2018.tar.bz2It's ~42 MB and labeled 2010-2018, so it's not just a couple of random days. I can't vouch for the exact channel layout from here, but given the rest of that directory (bitcoin-dev, mailing list mirrors, etc.) it's very likely to contain the #bitcoin-dev material you're after, or at least a huge chunk of it.
Second, if you really want from the original source, the old bitcoinstats logs are partially preserved on the Wayback Machine. A direct example:
https://web.archive.org/web/20131201200415/http://bitcoinstats.com/irc/bitcoin-dev/logs/2011/12/31 is a full day's transcript.
The Princeton CITP folks even wrote about scraping current vs Wayback versions and confirmed that a lot of days are archived there, though there are some holes and inconsistencies.
If the diyhpl tarball somehow doesn't pan out, the "nuclear option" is to script against Wayback's CDX API for
https://bitcoinstats.com/irc/bitcoin-dev/logs/* and reconstruct day-by-day from the snapshots.
For FLOSSmole: you already saw Megan's note that #bitcoin-dev logs were imported into their MySQL instance and that they deliberately didn't re-host the flat files because those were supposed to stay at the original sites.
Given that the MySQL box and account process are now effectively closed, I wouldn't bet on getting a SQL dump out of them unless one of the maintainers surfaces.
It's annoying that the "official" preservation story for such an important channel boils down to "some hackers' home dirs and the Wayback Machine", but that's pretty on brand for Bitcoin history.