[gax] wrap non-retryable RPCs in the same retryable callable machinery

[noahdietz]

Today, those methods that are configured at the client-level to be non-retryable would not benefit from some ApiCallContext improvements, like supporting withRetrySettings and withRetryableCodes added in #1238, and a proposal to add a new logical timeout option. This is because the callable for such methods is not wrapped in the same context-aware machinery that retryable methods are enabling such features.

We should consider wrapping the non-retryable RPCs in the same callable machinery that would enable retries. Since the retry settings would prevent retries, there shouldn't be a behavioral issue there.

[noahdietz]

cc @olavloite

[noahdietz]

Specifically, we would remove the following conditions

gax-java/gax/src/main/java/com/google/api/gax/rpc/Callables.java

Line 59 in 3fe1db9

if (areRetriesDisabled(callSettings.getRetryableCodes(), callSettings.getRetrySettings())) {

gax-java/gax/src/main/java/com/google/api/gax/rpc/Callables.java

Line 84 in 3fe1db9

if (areRetriesDisabled(callSettings.getRetryableCodes(), callSettings.getRetrySettings())) {

We would also need to ensure that the first attempt for non-retryable RPCs uses the proper timeout setting, as documented in those conditional statements.

[noahdietz]

@miraleung @vam-google WDYT?

[noahdietz]

Opened #1328 to address this if we think it is worth it.