Use chunk_id to deduplicate chunks in codecompanion tool

[Davidyz]

• Refactor the CLI so that the chunk id is returned in the results of the query subcommand
• Implement ID filtering when chunk_mode=true for codecompanion tool

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 99.49%. Comparing base (fc4391b) to head (0150ba0).
Report is 1 commits behind head on main.

Additional details and impacted files

@@           Coverage Diff           @@
##             main     #183   +/-   ##
=======================================
  Coverage   99.49%   99.49%           
=======================================
  Files          21       21           
  Lines        1589     1589           
=======================================
  Hits         1581     1581           
  Misses          8        8           

☔ 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.

[Davidyz]

I'm noticing that the refs with opts.invisible=false are being removed for every turn. I've narrowed it down to the chat:check_references() method using the content of the chat buffer to obtain a list of refs to be removed. I thought about patching this method so that it skips this check on invisible references, but I believe this'll make it difficult to delete invisible refs. @olimorris any suggestions?

EDIT: Would I be better off maintaining an in-house data structure to store the added chunks/documents?