try to figure out what breaks cached builds

[bcressey]

Issue number:
N/A

Description of changes:
Not for merging - trying to see why cargo is unhappy reusing builds from the cache.

Testing done:
😬

Terms of contribution:

By submitting this pull request, I agree that this contribution is dual-licensed under the terms of both the Apache License, version 2.0, and the MIT license.

[bcressey]

There are at least two problems here.

The first is that the fingerprints in the cache don't match the fingerprints that cargo generates, and some files seem to be missing:

# from cache
target/x86_64/debug/.fingerprint/microcode-203e868d1235a310
target/x86_64/debug/.fingerprint/microcode-3d5c23d8c6a2f178
target/x86_64/debug/.fingerprint/microcode-82b3ac875abbc53a

# cargo output
target/x86_64/debug/.fingerprint/microcode-203e868d1235a310/build-script-build-script-build
target/x86_64/debug/.fingerprint/microcode-06259485b1fc1ba5/run-build-script-build-script-build
target/x86_64/debug/.fingerprint/microcode-79bfebf693197603/lib-microcode

The second is that the mtime is reset for all files to the time of the clone:

  File: packages/microcode/latest-upstream-tags.sh
  Size: 355       	Blocks: 8          IO Block: 4096   regular file
Device: 801h/2049d	Inode: 9300713     Links: 1
Access: (0755/-rwxr-xr-x)  Uid: ( 1000/  runner)   Gid: ( 1000/  runner)
Access: 2024-08-02 21:43:03.524858363 +0000
Modify: 2024-08-02 21:43:03.524858363 +0000
Change: 2024-08-02 21:43:03.524858363 +0000
 Birth: 2024-08-02 21:43:03.524858363 +0000