Add /pr agent skill and refresh PR template

[ktehranchi]

Changes proposed in this Pull Request

Adds a repo-committed Claude Code skill, .claude/skills/pr/SKILL.md, invoked as /pr, that guides agents through opening a PR to this repository:

• Base branch: develop always (never master), with an exception for branches cut from an active integration branch (e.g. v1-epic), detected via merge-bases.
• Push remote: upstream when the invoking user has push permission, otherwise their fork with a cross-repo PR — so the same skill serves maintainers and external contributors.
• Pre-PR gauntlet: clean tree, up-to-date with base, pre-commit, and Tier A (pytest -m fast, mirroring CI's fast-tests job). For changes that can alter network artifacts, the agent prompts the user to run Tier B (pytest -m integration) — encouraged, not required.
• PR body: fills the template, adds a Testing section with tier evidence, and for larger changes an Equivalence section citing harness results / deltas-ledger rows — or an explicit statement that no run exists.
• Draft when evidence is missing on a larger change; watches the fast-tests check after opening.

Also refreshes .github/pull_request_template.md: checklist paths had gone stale (envs/environment.yaml, config.default.yaml, doc/configtables/*.csv → workflow/envs/environment.yaml, workflow/repo_data/config/config.default.yaml, docs/source/configtables/*.csv) and adds a Testing section the skill fills in.

Note: the skill references the tiered pytest markers (fast/integration) that land on develop with #775; until then Tier A resolves against the existing unit suite once that merges.

Testing

• pre-commit run --files <changed files>: all hooks passed.
• Tier A (pytest -m fast): not run — the tier markers/pytest config arrive with v1-epic: simplify-early pipeline with results-equivalence guarantee #775; this PR is docs/skill-only (no Python or workflow changes).
• Tier B / equivalence: not applicable (no artifact-affecting changes).

Checklist

• I tested my contribution locally and it seems to work fine.
• pre-commit and pytest -m fast pass locally.
• Code and workflow changes are sufficiently documented.
• Changed dependencies are added to workflow/envs/environment.yaml. (n/a)
• Changes in configuration options are added in workflow/repo_data/config/config.default.yaml. (n/a)
• Changes in configuration options are also documented in docs/source/configtables/*.csv. (n/a)

🤖 Generated with Claude Code