Cirq 1.5.x to Cirq 1.6.1

[psamanoelton]

Summary

This PR updates TensorFlow Quantum to support the current Cirq release line while preserving Python 3.10 support.

The resulting support matrix is:

• Python 3.10 -> Cirq 1.5.0
• Python 3.11-3.12 -> Cirq 1.6.1

This keeps TFQ compatible with TensorFlow 2.19.1 across Python 3.10-3.12 while still moving newer Python versions onto the current Cirq release line.

What changed

• Added Python-version-specific Cirq pins:
  • cirq-core==1.5.0 / cirq-google==1.5.0 for Python <3.11
  • cirq-core==1.6.1 / cirq-google==1.6.1 for Python >=3.11
• Updated release/setup.py metadata to support Python >=3.10,<3.13
• Updated configure.sh and release/build_pip_package.sh to allow Python 3.10, 3.11, and 3.12
• Added version-specific lockfiles:
  • requirements_lock_3_10.txt
  • requirements_lock_3_11.txt
  • requirements_lock_3_12.txt
• Updated lockfile generation to produce per-Python-version requirements locks
• Added helper scripts/rules to select the correct lockfile for the active interpreter in local scripts and Bazel
• Updated local Docker helpers to provide dedicated Python 3.10, 3.11, and 3.12 environments
• Updated docs to describe the split support model and version-aware dependency installation flow

Why this approach

Cirq 1.6.x requires Python 3.11 or newer, but TensorFlow 2.19.1 still supports Python 3.10.

Instead of dropping Python 3.10 entirely, this PR keeps support by splitting the dependency track:

• Python 3.10 stays on Cirq 1.5.0
• Python 3.11-3.12 move to Cirq 1.6.1

This preserves compatibility for existing Python 3.10 users while still allowing newer supported Python versions to use the current Cirq release line.

Python support after this PR

• Python 3.10: supported
• Python 3.11: supported
• Python 3.12: supported
• Python 3.13: not supported

Python 3.13 is still excluded because TensorFlow 2.19.1 does not publish Python 3.13 wheels.

Validation

Validated successfully with local Docker-based workflows for:

• Python 3.10
• Python 3.11
• Python 3.12

Validation included:

• dependency resolution using the version-specific lockfile
• TFQ source build
• wheel build
• wheel install
• runtime smoke test
• full test flow with the updated Docker helpers

Notes

• TFQ currently builds against CPU only in this flow
• configure.sh still defaults to CPU automatically
• GPU support was not re-enabled in this PR
• requirements.txt remains the primary dev lock, and version-specific lockfiles are used for Python-specific validation/build flows

[gemini-code-assist]

Code Review

This pull request drops support for Python 3.10, raising the minimum supported version to Python 3.11, and updates various dependencies including Cirq to 1.6.1 and NumPy to 2.1. Feedback on the changes points out that while Python 3.13 is explicitly unsupported, the version checks in configure.sh and release/build_pip_package.sh only enforce a lower bound of >= 3.11. It is recommended to add an upper bound check of < 3.13 to prevent build or installation failures on newer Python versions.