[Story]: JetBrains AI Analytics, Amazon Q and Tabnine connectors

## Summary
Build connectors for JetBrains AI Analytics, Amazon Q, and Tabnine using the same connector pattern as the Copilot/Cursor connectors. JetBrains uniquely exposes acceptance rate broken out per backing agent (Junie / Claude Code / Codex), which the normalizer must be able to represent. Amazon Q's acceptance metric excludes suggestions dismissed by continued typing — a different denominator than other vendors that the normalizer (story 6) must explicitly handle, not silently average away. Tabnine is the only surveyed vendor supporting a private/on-prem installation, which affects how its connector authenticates and where it can be deployed.

## Intent / Source of truth
Each vendor's own documented analytics/admin API. 2026-08-13 backlog consolidation, vendor capability survey.

## Scope
- [ ] JetBrains AI Analytics connector, preserving the per-backing-agent (Junie/Claude Code/Codex) breakdown
- [ ] Amazon Q connector, with its "excludes dismissed-by-continued-typing" denominator explicitly documented and tagged in the schema mapping
- [ ] Tabnine connector, supporting both SaaS and private/on-prem deployment targets
- [ ] Map all three into the schema from story 1

## Out of scope
- Building a unified UI across all three in this story (that's story 6, the normalizer)
- Any JetBrains IDE feature beyond AI Analytics

## Verification
Live pull from each of the three vendors' APIs showing real data, with the JetBrains per-agent breakdown and the Amazon Q denominator caveat both visibly represented in the ingested data (not flattened away).

## Risk assessment
Silently treating Amazon Q's different denominator as equivalent to other vendors' would corrupt cross-vendor comparisons — this must be caught here, not deferred to the normalizer to guess at.

## AI Usage Declaration
Drafted with AI assistance during the 2026-08-13 cross-repo backlog consolidation. A human owns intent, verification and consequences; acceptance criteria must be verified against the live system before this is closed.