Fixer: use HTML URL for issue fetch (avoids API rate limit 403s)

GitHub API has 60 req/hr unauthenticated limit per IP. Shared GHA
runners hit this easily. HTML page fetch always works for public repos.
Fall back to API URL only if HTML fails.

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

Author: Dan Moseley

.github/workflows/agentic-fix-issue.md

@@ -94,10 +94,11 @@ You are running in a personal fork of dotnet/runtime. The repo is checked out an
 ## Phase 0-1: Read Issue + Validate (1 TURN)
 
 **Do EVERYTHING below in a SINGLE turn — batch all tool calls in parallel:**
-- `web-fetch: https://api.github.com/repos/${{ inputs.upstream_repo }}/issues/${{ inputs.issue_number }}`
-- `web-fetch: https://api.github.com/repos/${{ inputs.upstream_repo }}/issues/${{ inputs.issue_number }}/comments`
+- `web-fetch: https://github.com/${{ inputs.upstream_repo }}/issues/${{ inputs.issue_number }}` (HTML — always works for public repos)
 - `bash: git log --oneline --all | grep ${{ inputs.issue_number }} || true`
 
+If web-fetch returns 403/429, retry with the API URL: `https://api.github.com/repos/${{ inputs.upstream_repo }}/issues/${{ inputs.issue_number }}`
+
 Do NOT use `gh issue view` CLI. Do NOT delegate to sub-agents.
 
 **From the fetched data, validate inline (no extra turn):**