Prefer RateLimitError on 429 with rate-limit body

[juanmanuelramallo]

What this does

Some providers include token-budget context in their HTTP 429 rate-limit bodies. Anthropic, for example, returns messages like "...rate limit of 5,000,000 input tokens per minute...". The "input tokens" substring matches the existing /input[_\s-]?token/i entry in CONTEXT_LENGTH_PATTERNS, so the current 429 branch classifies these as ContextLengthExceededError — making it impossible for callers to tell a token-quota rate-limit (retriable with backoff) apart from a true context-window overflow (non-retriable).

On 429, prefer RateLimitError whenever the body carries an explicit rate-limit signal ("rate limit"), and only fall back to ContextLengthExceededError when the message is unambiguously context-length-shaped. A single regex is sufficient — the Anthropic body and other documented rate-limit responses use the literal phrase "rate limit".

Preserves the existing "429 + Request too large for model" → ContextLengthExceededError behavior (that phrase has no rate-limit signal). All other status-code branches and the 400 branch are unchanged.

Type of change

• Bug fix
• New feature
• Breaking change
• Documentation
• Performance improvement

Scope check

• I read the Contributing Guide
• This aligns with RubyLLM's focus on LLM communication
• This isn't application-specific logic that belongs in user code
• This benefits most users, not just my specific use case

Required for new features

• I opened an issue before writing code and received maintainer approval
• Linked issue: #___

PRs for new features or enhancements without a prior approved issue will be closed.

Quality check

• I ran overcommit --install and all hooks pass
• I tested my changes thoroughly
  • For provider changes: Re-recorded VCR cassettes with bundle exec rake vcr:record[provider_name]
  • All tests pass: bundle exec rspec
• I updated documentation if needed
• I didn't modify auto-generated files manually (models.json, aliases.json)

AI-generated code

• I used AI tools to help write this code
• I have reviewed and understand all generated code (required if above is checked)

API changes

• Breaking change
• New public methods/classes
• Changed method signatures
• No API changes

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 87.22%. Comparing base (5bdda1a) to head (aa2a77f).

Additional details and impacted files

@@           Coverage Diff           @@
##             main     #775   +/-   ##
=======================================
  Coverage   87.21%   87.22%           
=======================================
  Files         121      121           
  Lines        5703     5707    +4     
  Branches     1442     1443    +1     
=======================================
+ Hits         4974     4978    +4     
  Misses        729      729           

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[juanmanuelramallo]

@crmne does it make sense to merge this?