docs(example): close out INFRA-TASKS with summary and 4 follow-up items

Adds a closing remark (23 Feb 2026) summarising the final state of the
infospace: 988 entities, 985 evaluations, 823 L2 classifications, 15 L3
relations, viability 6/6 PASS.

New open tasks 20–23:
  20. Complete L2 classification batch (165 entities blocked on credits)
  21. Run classify-links for 58 Relation-type entities
  22. Refresh stale metrics-report.md narrative
  23. Smoke-test the graph command end-to-end

Also committed: history.py fix — write_metrics_file now preserves
non-float metric values (type_distribution dict) instead of crashing
on round().

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

examples/infospace-with-history/INFRA-TASKS.md

@@ -573,3 +573,97 @@ Collection evaluation (85 entities, 7 chapters):
   Consistency:  0 cycles, 5 undefined deps, 0 contradictions
   Granularity:  entropy 1.42, 1 dispensable, 2 merge candidates
 ```
+
+---
+
+## Closing Remark (23 Feb 2026)
+
+All 19 original infrastructure tasks have been resolved (or intentionally
+deferred for scope reasons — tasks 4 and 7). The infospace now has:
+
+- **988 entities** extracted from 7 of 35 WoN chapters
+- **985 per-entity LLM evaluations** (5 dimensions, mean 3.96/5.0)
+- **823 L2 type × VSM classifications** (29/30 matrix cells occupied)
+- **15 L3 relation triplets** with 6 feedback loops detected
+- **Viability dashboard: 6/6 PASS**
+
+The tooling built here generalises beyond this example. The `markitect
+infospace` CLI family — `classify`, `evaluate`, `check`, `relations`,
+`graph`, `viability` — is ready to use on any new infospace.
+
+The following tasks remain open as of this writing:
+
+---
+
+## 20. Complete L2 classification batch — OPEN
+
+**Issue:** 165 of 988 entities remain unclassified. The batch was
+interrupted by OpenRouter credit exhaustion (HTTP 402) near the end of
+the alphabet (roughly `w–z` entities).
+
+**Impact:** `classify-summary` reports 823/988 entities; the type × VSM
+matrix has one empty cell (Institution/S4) that may fill once the
+remaining entities are classified.
+
+**Fix:** Run `markitect infospace classify --provider openrouter` (or
+`--provider gemini --rpm 10` after daily quota reset). Incremental skip
+handles the 823 already classified automatically.
+
+---
+
+## 21. Run classify-links for Relation-type entities — OPEN
+
+**Depends on:** Task 20 (full classification batch).
+
+**Issue:** 58 entities were classified as type `Relation`. These are
+structural connectors (e.g. "Rent determined by Price") that name two
+endpoints and a mechanism. The `classify-links` command captures this
+endpoint data (SUBJECT, OBJECT, MECHANISM fields), enriching the
+classification files and enabling the `entities --by-type` display
+to show the relation graph inline.
+
+**Impact:** Without `classify-links`, Relation-type entities are displayed
+without their linking data. The `graph` command cannot draw edges through
+Relation-typed entities as intermediaries.
+
+**Fix:** `markitect infospace classify-links --provider openrouter`
+(or gemini). Enriches existing classification files in-place; safe to
+re-run.
+
+---
+
+## 22. Refresh stale metrics report — OPEN
+
+**Issue (original task 6, still open):** `output/metrics/metrics-report.md`
+is a narrative report generated by the old pipeline. It predates the
+structured `metrics.yaml` / `history.yaml` system and reflects only
+the 7-chapter partial corpus without LLM evaluation scores.
+
+**Impact:** The narrative report is misleading — it shows outdated numbers.
+Readers may trust it over the authoritative `metrics.yaml`.
+
+**Fix:** Either regenerate via `markitect infospace process --post-batch`
+(if the `assess-metrics` template is updated to reference `metrics.yaml`),
+or archive the old report with a deprecation notice pointing to
+`output/metrics/metrics.yaml`.
+
+---
+
+## 23. Smoke-test the graph command end-to-end — OPEN
+
+**Issue:** `markitect infospace graph` was implemented and committed
+(S2 graph export module) but has not been exercised against real
+classified data from this example.
+
+**Impact:** The command may have edge cases when combining sparse
+classifications (823/988 entities) with the 15 seed relation files.
+Filters (`--type`, `--vsm`, `--loops`, `--entity`) and both output
+formats (Mermaid, DOT) should be verified.
+
+**Fix:** Run a few representative invocations:
+```
+markitect infospace graph --format mermaid --color-by vsm --loops
+markitect infospace graph --type Relation --format dot
+markitect infospace graph --entity division_of_labour
+```
+Fix any rendering issues found.