[VesterlundCoder]

Summary

Fixes the P0 "umbrella CLI reports unverifiable successes" finding from the 2026-10-24 release review and prepares westquant 0.1.1.

• westquant compare printed verified = yes for each Qiskit baseline and Verification: EXACT for the WestQuant result without running any check. Each row now shows the status that QiskitEquivalenceVerifier actually returned (exact, invalid or unknown). Reductions are only printed when the selected candidate is verified.
• compare -f pytket|pennylane|pulser returned exit code 0 with "coming soon"; it now prints "not implemented" to stderr and exits 2.
• -o 0 was silently replaced by level 2.
• version and doctor print the installed __version__ instead of a hard-coded string.
• README: removed from westquant import optimize (never exported) and the synthetic numbers shown as command output. It now documents the real DeterministicSearchEngine API and shows actual output for a 5-qubit QFT. That output shows no improvement over opt-2/3 for this circuit.
• The qiskit and all extras, and the test extra, require westquant-qiskit>=0.1.2, whose stage plugins work on Qiskit 2.x. This PR's CI can only pass once westquant-qiskit 0.1.2 is on PyPI.
• CI: visible nightly failures, and publish.yml is gated on tests.yml plus a tag/version check.
• Version 0.1.1. The review proposed 0.1.0a3, but 0.1.0a3 and 0.1.0 already exist on PyPI.

Test plan

• New tests: exact only when the verifier says so, unknown and no reductions for measured circuits, exit 2 for unimplemented frameworks, README imports must exist. 7 tests fail against the old code, 24 pass now
• wheel + sdist built, twine check --strict, installed with westquant-qiskit 0.1.2 and westquant-qcsc 0.2.0 in clean venvs, pip check OK, westquant resolves to this package
• Tests workflow green after westquant-qiskit 0.1.2 is published

Generated with Devin