Consider using actions/cache for per repo caching

[bryanhonof]

Describe the solution you'd like

Consider adding the actions/cache action for caching between CI runs. Something in the same spirit of the, see below. If I understand correctly, this will cause Bazel commands to first check if it can find a cache hit from the CI runner, before asking BuildBuddy to see if it has the results cached.

https://github.com/buildbuddy-io/buildbuddy/blob/ddb30ab25b6d1eb952eceb524cd2a754ae2d67c7/.bazelrc#L47-L56

https://github.com/buildbuddy-io/buildbuddy/blob/ddb30ab25b6d1eb952eceb524cd2a754ae2d67c7/.github/workflows/main.yaml#L20-L24

Describe alternatives you've considered

• Not adding the runner cache
• Depending on the remote cache, which will de done anyways

Additional context

Related to #243.

[avdv]

Yes, that could be useful. We are doing the same in the rules_haskell repository already:

https://github.com/tweag/rules_haskell/blob/4d9d65fd12dde216e3f3b9b13258368318be699a/.bazelrc#L59

https://github.com/tweag/rules_haskell/blob/4d9d65fd12dde216e3f3b9b13258368318be699a/.github/workflows/workflow.yaml#L17-L21