fanquake

> why not bumping miniupnpc sources I haven't tested them, and don't currently plan to. The 2.2.3 tag is also broken out of the box.

> why not bumping miniupnpc sources There's also no release/tag that would include my patch. So are you suggesting we move to using the master branch of miniupnpc? I doubt...

> I'd slightly prefer a full list of accepted headers in a separate file, instead of a list of commands in the makefile having -r directives which could come to...

Sorry for all the force push noise, this now (CI passing) seems to be in a working state for all HOSTS. Boost Process is a mess, and is the primary...

Rebased past #22953. Updated the PR description. Might split out one other related change.

Rebased on master, #25803, and a commit from 25696 that updates Boost to 1.80.0. Also updated to remove boost/detail/winapi, which is a duplicate of boost/winapi, and no-longer needed.

Split (and extended) part of this out into #26148 after discussion with @theuni. Also now rebased for Boost 1.80.0, and based on #25465. Please review either of those two PRs...

Updated to include the BOOST_ASIO_STANDALONE commit, and drop some additional dirs from depends, including filesystem.

[Similar to #25111. I'm still ~0 on this change](https://github.com/bitcoin/bitcoin/pull/25111#pullrequestreview-970602390), but this approach is an improvement. Given #24907, I don't think we should be leaning back into more Boost code usage....

> Bumping the Boost version in depends should probably be its own PR? See #25873.