tar driver: add --cache-dir to reuse the extracted filesystem across tests

[yuriipolishchuk]

Fixes #86. Also resolves the slowness reported in #326.

The tar driver creates a new driver per test, and each one extracts the full image filesystem into a fresh random temp dir. The cache plumbing to avoid this already exists (getExtractPathForName + the DirIsEmpty reuse check in GetFileSystemForImage) but nothing wires it up: GetImageForName hardcodes an empty cacheDir.

This adds a --cache-dir flag that threads through to the tar driver. The extraction path is keyed by image name/digest under the cache dir, so different images sharing a dir cannot reuse each other's filesystem. Combine with --save so per-test Destroy() keeps the extraction. Default behavior (no flag) is unchanged.

Benchmark (634MB image, 19 fileExistenceTests, --driver tar):

	wall time	extractions
without --cache-dir	3m13s	19
with --save --cache-dir	14s	1 (+18 cache reuses)

Identical test verdicts in both runs.

[gemini-code-assist]

Caution

The consumer version of Gemini Code Assist on GitHub has been sunset. All code review activity has officially ceased.

[google-cla]

Thanks for your pull request! It looks like this may be your first contribution to a Google open source project. Before we can look at your pull request, you'll need to sign a Contributor License Agreement (CLA).

View this failed invocation of the CLA check for more information.

For the most up to date status, view the checks section at the bottom of the pull request.