[CI] Dry x86-64-v3 builds/default images to use glibc

[wileyj]

ref:
#4364
#4383

This adjusts the changes introduced by @jbencin that were merged in 4364 for jemalloc, and adding an additional cpu arch x86-64-v3.

• removes the duplicated workflow to build x86-64-v3 binaries (current workflow was DRY'ed to build them via a matrix with some specific excludes)
• fixes the dockerfiles currently in ./build-scripts to set the RUSTFLAGS env var if TARGET_CPU build-arg is provided
• adjusts the conditional compilation flags in several files to exclude the target_os (required when cross-compiling - target_env is only application when building on a specific target triple. for github actions, the excude of target_env = msvc will not match to true since default is ubuntu vm).
• changes the default release docker image to use debian glibc as opposed to alpine (alpine is still built, but is tagged as <release version>-alpine
• builds the docker source image (which is triggered on any PR) with TARGET_CPU provided as x86-64-v3
• release docker images are changed to support the following cpu types: x86-64-v3, arm64, armv7. if desired, i think we can add support for x86-64 alongside x86-64-v3

sample actions runs from my fork (testing these chagnes on a release):
https://github.com/wileyj/stacks-blockchain/actions/runs/8087234345/job/22100013368
release: https://github.com/wileyj/stacks-blockchain/releases/tag/3.0.0.0-rc1
images:
debian release image
alpine release image

sample action for the source docker image:
https://github.com/wileyj/stacks-blockchain/actions/runs/8101666085/job/22142408056
resulting image:
debian x86-64-v3 image from source

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 77.43%. Comparing base (04e5eb2) to head (14c3949).

Additional details and impacted files

@@            Coverage Diff             @@
##             next    #4460      +/-   ##
==========================================
- Coverage   83.23%   77.43%   -5.81%     
==========================================
  Files         451      451              
  Lines      325724   325724              
  Branches      323      323              
==========================================
- Hits       271111   252216   -18895     
- Misses      54605    73500   +18895     
  Partials        8        8              

Files	Coverage Δ
stackslib/src/main.rs	0.06% <ø> (ø)
testnet/stacks-node/src/main.rs	0.26% <ø> (ø)

... and 166 files with indirect coverage changes

---

Continue to review full report in Codecov by Sentry.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 04e5eb2...14c3949. Read the comment docs.

[obycode]

Looks good, there was just one comment that showed up in a bunch of docker files that I don't think was intended.

[jbencin]

Looks good, it's an improvement over what I had done

[wileyj]

@jbencin @obycode i could also use some help undertanding the conflict i have here in clarity/Cargo.toml:

<<<<<<< ci/dry_x64_builds
[target.'cfg(all(target_arch = "x86_64", not(any(target_os="windows"))))'.dependencies]
sha2-asm = "0.5.3"

=======
>>>>>>> next

is this dependency being removed entirely?

[obycode]

Right. Jeff was able to remove that dependency.

[obycode]

lgtm!

[zone117x]

LGTM. Kill x86-64-(legacy) imo -- can always add it back later if there is demand

[jbencin]

is this dependency being removed entirely?

Right. Jeff was able to remove that dependency.

See #4435, which has been merged. sha2-asm is used by sha2 and doesn't need to be included directly

[CharlieC3]

Great job on this!

[wileyj]

LGTM. Kill x86-64-(legacy) imo -- can always add it back later if there is demand

that's doable, but it raises an interesting question around naming:
currently i'm naming v3 compiled binaries as linux-glibc-x64-v3.zip

and the legacy build as linux-glibc-x64.zip

if we were to no longer build linux-glibc-x64.zip, i'd argue we should rename the v3 build as linux-glibc-x64.zip etc.

low chances that someone is trying to run this binary on a chip that is ~10 years old, i think defaulting to v3 is the right call here.

[wileyj]

LGTM. Kill x86-64-(legacy) imo -- can always add it back later if there is demand

killed. we'll now build the same number of binaries/images but x64 builds will default to v3.
i've added enough logic in the build steps too, so we can add in v2 or later variants as needed

[wileyj]

ping for second approval

[blockstack-devops]

This pull request has been automatically locked since there has not been any recent activity after it was closed. Please open a new issue for related bugs.