RPC timeouts and retry loops across services

The `google::cloud::bigtable::RPCRetryPolicy` classes have special member functions to setup a timeout:

https://github.com/googleapis/google-cloud-cpp/blob/81adbdabca2da794875fd567591fddc2cf0c8394/google/cloud/bigtable/rpc_retry_policy.h#L98-L101

https://github.com/googleapis/google-cloud-cpp/blob/81adbdabca2da794875fd567591fddc2cf0c8394/google/cloud/bigtable/rpc_retry_policy.cc#L53-L57

This is used in the synchronous retry loop, e.g.:

https://github.com/googleapis/google-cloud-cpp/blob/81adbdabca2da794875fd567591fddc2cf0c8394/google/cloud/bigtable/table.cc#L90-L95

But not in the asynchronous retry loop:

https://github.com/googleapis/google-cloud-cpp/blob/81adbdabca2da794875fd567591fddc2cf0c8394/google/cloud/internal/async_retry_unary_rpc.h#L145-L149

That asynchronous retry loop is used in spanner too, and the `RetryPolicy` for spanner does not have a `Setup()` member function. Furthermore, the same retry policy class is used in storage, where we would't want to introduce it (because that would create an unwanted dependency on gRPC).

We need a cleaner way to setup per-RPC timeouts across all services, with good support for REST, and we need to use that on the asynchronous and synchronous loops.

Start with google/cloud/bigtable/rpc_retry_policy.h and .cc, google/cloud/bigtable/table.cc, and google/cloud/internal/async_retry_unary_rpc.h. Compare the synchronous and asynchronous retry loops and the Bigtable and Spanner retry policies, then define what a shared per-RPC timeout mechanism must provide for REST without adding a gRPC dependency to Storage.

Written by the indexing model from the issue text.