Add get_sources timeout to runner config

What does this MR do?

Adds a get_sources_timeout runner config option that bounds the get_sources stage independently of the overall job timeout.

When the configured duration is exceeded, the job fails fast with Job failed: get sources timeout: exceeded timeout of <duration>. Falls back to the job timeout if unset, zero, or greater than the job timeout (with a warning logged for the latter two). Follows the same pattern as prepare_timeout.

Why was this MR needed?

Hung git clone/git fetch operations (network issues) currently consume the full job timeout, wasting runner concurrency slots and delaying feedback. get_sources_timeout lets administrators fail these jobs quickly without lowering the job timeout used by the actual work.

What's the best way to test this MR?

• Run go test ./common/ -run "TestGetGetSourcesTimeout|TestBuildGetSources" -count=1 -v
• For end-to-end verification, you can set get_sources_timeout to a short value in config.toml and trigger a job. The job log should end with Job failed: get sources timeout: exceeded timeout of <duration> and the failure attribution should match the existing job timeout behaviour otherwise.

What are the relevant issue numbers?

Implements #39426 (closed)