Add chart versioning policy

[bschwedler]

• Enforce <= 1.0.0 in release
• Add versioning policy

[lachlansimpson]

Thanks for picking this up, and for writing the convention down. The policy text
matches what you described on the issue, and routing it through
rstudio.description is a nice touch: one edit covers all six READMEs.

A few things before it leaves draft:

CONTRIBUTING.md / CLAUDE.md. The policy landed in consumer-facing docs,
but the person choosing minor vs. patch is reading CONTRIBUTING.md, which
today only says that a bump is required, not which position. Could the same
text go there? While it's open, two lines in it are stale: the bump rule no
longer covers docs-only changes (#839), and the file:// coupling was replaced
by pinned deps (#601).

The two non-product charts. "Product versions belong in appVersion" now
renders in the rstudio-library and rstudio-launcher-rbac READMEs, where
appVersion is SemVer and tracks no product. A sentence scoping that line to
the three product charts would cover it.

The pre-1.0.0 guard. I read it as an interlock, not a stance on graduation:
the release job hard-codes a patch bump, so failing early beats publishing a
wrong patch. If that's the intent, could the error message say so? As written it
fails a state the policy on the same PR describes as valid, which will confuse
whoever hits it.

The CI check (proposal 2). Is it planned as a follow-up?
check-news-entries.sh already has both versions and the NEWS entry in hand, so
the BREAKING/DEPRECATED match is a small addition. It's the piece that would
have caught 0.7.2, 0.20.1, 0.20.10, and workbench 0.21.1. Happy for it to be
separate, just want to know it isn't dropped.

Two tiny ones: the PR body says "<= 1.0.0" where the code enforces "< 1.0.0",
and the **/*.quarto_ipynb gitignore entry looks unrelated.

[jforest]

Forever? or is this just a stop gap for now to prevent third parties from bumping us to 1.x against our will?

[bschwedler]

Yes, this is just a stop-gap until we reach 1.0. I can add a comment or file an issue to remove this later.

[ianpittwood]

LGTM