Update smoke-ci haiku cache-memory behavior and issue body count

[Copilot]

Summary

Update the smoke-ci workflow instructions so each run:

• counts how many haiku records already exist in cache-memory
• writes a new haiku JSON record to cache-memory
• includes the run URL, generated haiku text, and existing haiku count in the created issue body

Validation

• Ran make recompile
• Ran make agent-finish (fails due to pre-existing testifylint issues in pkg/agentdrain/spec_test.go unrelated to this change)
• Ran parallel validation (Code Review + CodeQL) after final update

[Copilot]

Pull request overview

Updates the Smoke CI workflow instructions to persist and report haiku-related cache state across runs.

Changes:

• Add a cache-memory inspection step to compute existing_haiku_count before writing a new record.
• Persist a new per-run haiku JSON record using a filesystem-safe timestamp filename.
• Expand the created issue body requirements to include the run URL, haiku text, and existing record count.

Show a summary per file

File	Description
.github/workflows/smoke-ci.md	Adjusts the prescribed tool-call sequence to count existing haiku cache entries, write a new haiku record, and include the count + haiku in the issue body.

Copilot's findings

Tip

Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

• Files reviewed: 1/1 changed files
• Comments generated: 2

[Copilot]

Step 1 assumes /tmp/gh-aw/cache-memory/smoke-ci-haiku/ already exists. On a first run (or after cache eviction), inspecting/listing this directory may fail and break the workflow. Please clarify that if the directory is missing, existing_haiku_count must be set to 0 and the directory should be created before writing the new record.

Suggested change

	1. Use `cache-memory` to inspect `/tmp/gh-aw/cache-memory/smoke-ci-haiku/`, count how many haiku records already exist, then save that count as `existing_haiku_count`.
	2. Create a new haiku for this run and use `cache-memory` to save it as a JSON record in `/tmp/gh-aw/cache-memory/smoke-ci-haiku/` with a filesystem-safe timestamp filename in `YYYY-MM-DD-HH-MM-SS-sss` format (no `:`).
	1. Use `cache-memory` to inspect `/tmp/gh-aw/cache-memory/smoke-ci-haiku/` and count how many haiku records already exist; if the directory is missing, set `existing_haiku_count` to `0`.
	2. Create a new haiku for this run and use `cache-memory` to save it as a JSON record in `/tmp/gh-aw/cache-memory/smoke-ci-haiku/`, creating that directory first if it does not already exist, with a filesystem-safe timestamp filename in `YYYY-MM-DD-HH-MM-SS-sss` format (no `:`).

[Copilot]

The new JSON record format is underspecified: it doesn’t say what fields must be included (e.g., run URL, haiku text, timestamp, existing count) or even whether the filename should end in .json. Without a defined schema/extension, future reads and the file-counting logic can become inconsistent. Please specify the required JSON keys and a consistent filename pattern (including extension).

Suggested change

	1. Use `cache-memory` to inspect `/tmp/gh-aw/cache-memory/smoke-ci-haiku/`, count how many haiku records already exist, then save that count as `existing_haiku_count`.
	2. Create a new haiku for this run and use `cache-memory` to save it as a JSON record in `/tmp/gh-aw/cache-memory/smoke-ci-haiku/` with a filesystem-safe timestamp filename in `YYYY-MM-DD-HH-MM-SS-sss` format (no `:`).
	1. Use `cache-memory` to inspect `/tmp/gh-aw/cache-memory/smoke-ci-haiku/`, count how many existing haiku record files match `*.json`, then save that count as `existing_haiku_count`.
	2. Create a new haiku for this run and use `cache-memory` to save it in `/tmp/gh-aw/cache-memory/smoke-ci-haiku/` as a JSON object with exactly these required keys: `run_url`, `haiku_text`, `timestamp`, and `existing_haiku_count`. Use the consistent filename pattern `YYYY-MM-DD-HH-MM-SS-sss.json` (filesystem-safe, no `:`).