Summary

Add a --no-venv flag (and a corresponding AnalysisOptions field) that tells codeanalyzer-python to skip virtualenv creation / dependency installation and run analysis against the current / ambient Python environment instead.

Motivation

Today the analyzer provisions a virtualenv (and installs the target project's dependencies) so Jedi can resolve imports. In several contexts that step is unwanted:

• CI / containers where the project's dependencies are already installed in the active environment — the extra venv + install is redundant and slow.
• Ephemeral / sandboxed runs where network installs are disallowed or undesirable.
• Speed / reproducibility — reusing the ambient interpreter avoids minutes of per-run setup.

Proposed behavior

• --no-venv CLI flag → AnalysisOptions(no_venv=True).
• When set, skip venv creation and dependency installation; point Jedi at the current sys.prefix / active interpreter.
• Document the tradeoff: import/call-resolution quality then depends on what is installed in the ambient environment.

Notes

This pairs naturally with library / in-process use from the CLDK SDK (PyCodeanalyzer) and with --emit neo4j, where the host environment is often already the project's environment and the venv bootstrap is pure overhead.