Implement new Bigtable timeout & retry settings

[sming]

Use Case Resolved: Heroic pipeline instability and poor user experience for Grafana

• I am a Heroic developer
• who wants to armour up Heroic so that we get fewer alerts
• so that we can focus on features

Plan

• look at the spotify-bigtable-client repo for "inspiration"
• don't implement a whole new layer of configuration. Focus on the 2 or 3 settings we're immediately interested in.

Notes for Reviewer

• only read requests are affected (i.e. Heroic API but not metric Ingestion)
• a maxMutateRowTimeout is implemented but will be deployed with the default setting value so as not to change Ingestion at all. This is done for development efficiency reasons i.e. it would have been foolish not to add support for it whilst I was changing the exact same files in the exact same places. Here is the heroic-gke PR that puts mutateRpcTimeoutMs back to 10 minutes.

[sming]

was temporary PR - deleting