Migrate to cmake

[luke-jr]

Bitcoin Core 29.x is switching to CMake instead of Autotools, which means everything touching the build system will need rewriting.

This includes:

• sys_leveldb: cmake: Add unsupported WITH_SYSTEM_LEVELDB option #101
• sys_libsecp256k1: luke-jr/bitcoin@ae1e6bb...luke-jr:bitcoin:sys_libsecp256k1
• bugfix_asm_pragmas: crypto/sha256: Use pragmas to enforce necessary intrinsics for GCC and Clang #102
• 2020mingwthrd (should be evaluated if we can just drop it entirely?): Restore linking to libmingwthrd #103
• boost_171_177_workarounds (no longer using boost::process, so unnecessary)
• hww_windows: __branch_29...luke-jr:bitcoin:hww_windows-29
• mempressure
• conf_append_cxxflags (already included in Core)
• ionice + ionice_win
• dsha256_power8
• tor_subprocess
• win_taskbar_progress
• macos_dmg
• svg_icon
• tbc_font
• knots_branding
• subproc_closefds (run_command: Close non-std fds when execing slave processes bitcoin/bitcoin#30756)
• restore_libconsensus

Additionally, this removes the major pre-generation step of the source tarball, so it may make sense to just require rsvg-tools and ImageMagick rather than try to pre-render the icons.

(If anyone would like to work on these, ping me to ensure you have the latest branches.)

[michaelfolkson]

What is the current status of Knots cmake builds? Still needs these final task items to be completed before Knots can be built with cmake?

[luke-jr]

Yes.

win_taskbar_progress is rebased, but needs testing still.

macos_dmg requires more than just CMake at this point; I had to drop it for 28.x in preference for signed/notarised binaries required for Apple to allow users to run the builds at all.

svg_icon and knots_branding are probably trivial (based on experience rebasing tbc_font) if ImageMagick and librsvg2 are made unconditional GUI dependencies.

restore_libconsensus may be difficult

[michaelfolkson]

Very much a cmake newbie but in the absence of getting help from one of the Core build system guys (hebasto etc) I or someone else could focus on getting up to speed with cmake. Definitely a full time project though (for someone).

[luke-jr]

I'm not sure it's that difficult... maybe full time for a week or two at most.

[michaelfolkson]

For someone with no prior cmake, build system experience? Ok, if you say so.

[luke-jr]

Maybe add a few weeks for no prior experience. But it makes more sense to have people familiar with things working on those areas, IMO.

[michaelfolkson]

Sure if someone with that skillset pops up

[luke-jr]

#133 is probably ready for review and testing