red-knot: source_text, line_index, and parsed_module queries

[MichaReiser]

Summary

This PR adds the most basic file-based queries that are needed for semantic analysis, linting and formatting:

• source_text: Reads the content of a file. Changing a file's revision markes the query result as stale. I excluded jupyter notebook support for now.
• line_index: Computes the line index for a file's source text.
• parsed_module: Parses the file's text to a python module, infering the file type from the file extension.

Test Plan

I added a few unit tests that demonstrate the basic functionality.

[github-actions]

ruff-ecosystem results

Linter (stable)

ℹ️ ecosystem check encountered linter errors. (no lint changes; 1 project error)

mlflow/mlflow (error)

Failed to clone mlflow/mlflow: error: RPC failed; curl 55 Failed sending HTTP2 data
error: 27308 bytes of body are still expected
fetch-pack: unexpected disconnect while reading sideband packet
fatal: early EOF
fatal: fetch-pack: invalid index-pack output

Linter (preview)

✅ ecosystem check detected no linter changes.

Formatter (stable)

✅ ecosystem check detected no format changes.

Formatter (preview)

✅ ecosystem check detected no format changes.

[carljm]

Does it make sense to assume a file is Python if it has no extension? Is that an existing Ruff behavior? That seems likely to lead to a lot of syntax/parsing errors. From a user perspective, I would expect to have to specially request via config if I want Ruff to treat an unknown-extension file as Python source.

[MichaReiser]

We'll need to filter out non-python files (according to the user's configuration) when walking the workspace so that we never end up calling parse on these files.

[carljm]

Tests also for stub file, vendored stub file, and unknown-extension file?

[MichaReiser]

I added a test for vendored paths. I prefer to not add a test for unknown extensions and stub files because I then end up re-testing PySourceType::from_extension. I rather want to have integration tests for different file types (We write a lot of integration tests in ruff and only few unit tests. What we have in these PRs is already way above the average number of unit tests)

[carljm]

Oh, and let's make sure to fix the clippy errors before merging this.