feat(model-profiles): plain-English summary for profile refresh PRs (#38218)

Automated model-profile refresh PRs (e.g. #38210) ship a static template
body, so a reviewer has to open *Files changed* and read large blocks of
generated data to learn what actually moved. Because the underlying
profile data is fully structured, we can describe the changes
deterministically — no LLM, no hallucination risk.

This adds a `langchain-profiles summarize` subcommand that compares the
working-tree `_profiles.py` files against a git ref and renders a
skimmable Markdown summary: models added (with a short capability
descriptor), models removed, and per-field capability changes
(context/output tokens, modalities, tool calling, reasoning, etc.),
grouped by provider and capped so huge refreshes stay readable. Profiles
are read with `ast.literal_eval` rather than imported, so the generated
data file is never executed.

Example output for a refresh that adds a model and bumps an output
limit:

```
## Summary of changes

**1 added · 0 removed · 1 changed** across 1 provider(s).

### openai

**➕ 1 added**
- `gpt-6-preview` — 1,000,000 ctx, 128,000 out, text+image+audio in, reasoning, tools

**✏️ 1 changed**
- `gpt-3.5-turbo`: max output tokens 4,096 → 16,384
```

Made by [Open
SWE](https://openswe.vercel.app/agents/9bcbf182-effc-ba9b-0df3-afac620ad152)

---------

Co-authored-by: open-swe[bot] <open-swe@users.noreply.github.com>

.github/workflows/_refresh_model_profiles.yml

@@ -167,6 +167,32 @@ jobs:
             exit 1
           fi
 
+      - name: "📝 Build PR body with change summary"
+        id: pr-body
+        env:
+          PROVIDERS_JSON: ${{ inputs.providers }}
+          PR_BODY: ${{ inputs.pr-body }}
+        run: |
+          # The refresh step modified the working tree without committing, so
+          # comparing against HEAD yields exactly the refresh's changes.
+          cli_dir="${GITHUB_WORKSPACE}/${{ steps.cli.outputs.dir }}"
+          body_file="${RUNNER_TEMP}/pr_body.md"
+          printf '%s\n\n' "${PR_BODY}" > "${body_file}"
+          # `summarize` builds the whole summary in memory and prints it once,
+          # so a failure exits non-zero before any stdout reaches the append —
+          # the body keeps only the static note, never a half-written summary.
+          if ! uv run --frozen --project "${cli_dir}" \
+            langchain-profiles summarize \
+            --providers "${PROVIDERS_JSON}" \
+            --base-ref HEAD \
+            --repo-root "${GITHUB_WORKSPACE}" >> "${body_file}"; then
+            echo "::warning::Could not generate change summary; see job log."
+            # Surface the degradation in the PR body too: the warning above only
+            # lands in the Actions log, which a PR reviewer won't see.
+            printf '\n> [!NOTE]\n> Automated change summary unavailable — see the workflow run log.\n' >> "${body_file}"
+          fi
+          echo "path=${body_file}" >> "$GITHUB_OUTPUT"
+
       - name: "🔑 Generate GitHub App token"
         id: app-token
         uses: actions/create-github-app-token@bcd2ba49218906704ab6c1aa796996da409d3eb1 # v3
@@ -182,7 +208,7 @@ jobs:
           branch: ${{ inputs.pr-branch }}
           commit-message: ${{ inputs.pr-title }}
           title: ${{ inputs.pr-title }}
-          body: ${{ inputs.pr-body }}
+          body-path: ${{ steps.pr-body.outputs.path }}
           labels: ${{ inputs.pr-labels }}
           add-paths: ${{ inputs.add-paths }}