Add pr-best-practices skill

[Thanhphan1147]

What this PR does

Add a skill that when called will analyze the changes between the first and final iteration of a PR authored by an LLM. The result is a set of good and bad example maintained in .github/instructions/best-practices.instructions.md. For example:

## Handling Jinja2 templates

Keep rendering logic in charm-state dataclasses or helper builders so templates stay declarative.
State properties that produce ACL conditions should return bare condition strings; the template is responsible for wrapping them in `acl` directive syntax.

<examples>

Checklist

• I followed the contributing guide
• I added or updated the documentation (if applicable)
• I updated docs/changelog.md with user-relevant changes
• I added a change artifact for user-relevant changes in docs/release-notes/artifacts. If no change artifact is necessary, I tagged the PR with the label no-release-note.
• I used AI to assist with preparing this PR
• I added or updated tests as needed (unit and integration)
• If integration test modules are used: I updated the workflow configuration
  (e.g., in .github/workflows/integration_tests.yaml, ensure the modules list is correct)
• If this PR involves a Grafana dashboard: I added a screenshot of the dashboard
• If this PR involves Terraform: terraform fmt passes and tflint reports no errors

[github-actions]

Test results for commit 9b28b7d

Test coverage for 9b28b7d

Name                               Stmts   Miss Branch BrPart  Cover   Missing
------------------------------------------------------------------------------
src/charm.py                          45      9      2      0    77%   65-91, 96-98
src/haproxy_spoe_auth_service.py      44     16      2      0    61%   56-64, 76-82, 93-117
src/state.py                          55     15      6      1    67%   64-66, 79, 125-146
------------------------------------------------------------------------------
TOTAL                                144     40     10      1    68%

Static code analysis report

Run started:2026-05-15 23:07:17.024833

Test results:
  No issues identified.

Code scanned:
  Total lines of code: 409
  Total lines skipped (#nosec): 1
  Total potential issues skipped due to specifically being disabled (e.g., #nosec BXXX): 1

Run metrics:
  Total issues (by severity):
  	Undefined: 0
  	Low: 0
  	Medium: 0
  	High: 0
  Total issues (by confidence):
  	Undefined: 0
  	Low: 0
  	Medium: 0
  	High: 0
Files skipped (0):

[srbouffard]

I wonder if we could have a way to split best practicies into multiple instructions to avoid having a single one that applies too all files.
My worry here is that as we accumulate more and more practices, we will front-load a LOT of context that might not be needed by every agent, every prompt.