i#4549: Finalize package building and docs deployment

[derekbruening]

Fixes a few issues with the script added in PR #4612:

• Splits the a64 and a32 builds into separate jobs, as package.cmake doesn't support both at once.
• Fixes minor issues.

Tested via a manual launch with filled-in version and build and it worked:

• https://github.com/DynamoRIO/dynamorio/actions/runs/424016479
  (longest build time: Windows took 29:15)
• https://github.com/DynamoRIO/dynamorio/releases/tag/release_8.0.18611-2
  All 5 files are present and with similar sizes to the last Travis+Appevyor-produced build, indicating they do contain Dr. Memory.
• https://dynamorio.org/dynamorio_docs/ has a new timestamp: successful docs deployment

Issue: #4549

[derekbruening]

Failure is histogram.offline in 64-bit: it's already on the list for 32-bit for #4621. I'll add it to the list separately (or maybe part of PR #4623).

[derekbruening]

I also updated https://github.com/DynamoRIO/dynamorio/wiki/New-Release#trigger-a-package-build

[abhinav92003]

I also updated https://github.com/DynamoRIO/dynamorio/wiki/New-Release#trigger-a-package-build

I'm not sure if anyone could trigger a manual build before, but is that acceptable? Do we need to somehow restrict this action to only admins?

[derekbruening]

I also updated https://github.com/DynamoRIO/dynamorio/wiki/New-Release#trigger-a-package-build

I'm not sure if anyone could trigger a manual build before, but is that acceptable? Do we need to somehow restrict this action to only admins?

I think it's ok to allow any regular committer who has write access to the repository to trigger a new posted build? Before any dev could trigger as well by going to the Travis manual trigger page using Github credentials.