[WIP] [v1.x] Create tool for building source archives

[josephevans]

Description

This PR introduces a script that automates creating a release source archive and handles some licensing issues by removing some files that were identified to be non-compliant, and introducing a Apache RAT license check in the source generation process to ensure all Apache MXNet code is compliant.

[mxnet-bot]

Hey @josephevans , Thanks for submitting the PR
All tests are already queued to run once. If tests fail, you can trigger one or more tests again with the following commands:

• To trigger all jobs: @mxnet-bot run ci [all]
• To trigger specific jobs: @mxnet-bot run ci [job1, job2]

---

CI supported jobs: [centos-gpu, unix-gpu, edge, clang, windows-cpu, unix-cpu, centos-cpu, miscellaneous, windows-gpu, website, sanity]

---

Note:
Only following 3 categories can trigger CI :PR Author, MXNet Committer, Jenkins Admin.
All CI tests must pass before the PR can be merged.

[Zha0q1]

@josephevans Does this same tool also work for master?

[josephevans]

@josephevans Does this same tool also work for master?

There will be some slight modifications for master (there's no R-package in master, which we remove in v1.x.)

[josephevans]

I was going to remove mx-theme references in the LICENSE file, but it looks like @szha already has them removed in #19873. Should we also remove references to mathjax (from mkldnn docs) in LICENSE since we will no longer be shipping it out in the source archive? @leezu

[leezu]

Should we also remove references to mathjax (from mkldnn docs) in LICENSE since we will no longer be shipping it out in the source archive?

One idea is to automatically remove lines fro LICENSE that match a line in tools/source-exclude-artifacts.txt?

[szha]

It would be great to add a check for all files listed in license and warn if any doesn't exist

[josephevans]

It would be great to add a check for all files listed in license and warn if any doesn't exist

Added this to the script. Found 2 items in the LICENSE file that are no longer in the source tree and removed them.

Perhaps this should be part of the normal sanity rat check CI pipeline so we catch the issues earlier instead of at release time. WDYT? But that will have to be part of a different PR.

[josephevans]

Should we also remove references to mathjax (from mkldnn docs) in LICENSE since we will no longer be shipping it out in the source archive?

One idea is to automatically remove lines fro LICENSE that match a line in tools/source-exclude-artifacts.txt?

OK, I updated to remove lines in LICENSE for stuff we remove the source archive.