Bundle official Qiskit skills and document agent task routing - #10

Bundle all four official Qiskit AI Skills so the lab can consult specialized guidance for Runtime migration, error mitigation, SQD eigensolver selection and fermionic simulation. Add task routing in AGENTS.md and document the skills, their uses and the official repository in the README. DAM's existing bundledSkills path discovers them alongside the project-specific qiskit-lab skill.

Vendor complete directories from Qiskit/skills commit 0888fb799b8a3972da65c2c0ea1965e27957111d, including registries, references and helper scripts. Preserve the Apache 2.0 license and upstream bytes, and record source paths and Git blob hashes in .agents/qiskit-skills.json. A packaging test exercises the actual tracked-file export and verifies that every upstream resource reaches the deployment bundle unchanged.

The SDK lockfile is unchanged. Optional fermionic/SQD packages remain task-specific dependencies. Existing DAM agents receive the skills through the documented checkout update; this PR does not deploy to a live agent. The project instructions retain the existing hardware execution, credential and reporting workflow.

Validation:

• uv sync --locked
• uv run --locked pytest -q --cov --cov-report=term-missing --cov-report=xml:coverage.xml — 192 passed.
• uv run --locked coverage report --include='*/job_tracking.py' --fail-under=100 — 100% line and branch coverage.
• uv run --locked ruff check src tests scripts/dam — passed.
• uv run --locked --group typing mypy — passed.
• uv build and shell syntax checks — passed.
• Verified all 35 upstream files against the pinned Git blob hashes. Upstream capabilities-table.md includes a trailing blank line; it is intentionally retained to preserve the exact source. Whitespace checks pass with blank-at-EOF checking excluded.
• The README still contains no Carbon references. IBM connectivity and live DAM deployment were not tested.