Pass test time estimate arg to stabilize smart flank without cached results

[RainNapper]

Context: #632

Ran into some local issues running all of the tests (stacktrace below[1]), but I think I covered most if not all of the places that this change touches.

Please leave comments on what I missed, or what I need to get a local build working. I can run unit tests in Android Studio, but hit some issues[2] running ./gradlew build

[1]

java.lang.VerifyError: Stack map does not match the one at exception handler 77
Exception Details:
  Location:
    com/fasterxml/jackson/databind/deser/std/StdDeserializer._parseDate(Lcom/fasterxml/jackson/core/JsonParser;Lcom/fasterxml/jackson/databind/DeserializationContext;)Ljava/util/Date; @77: astore
  Reason:
    Type 'com/fasterxml/jackson/core/JsonParseException' (current frame, stack[0]) is not assignable to 'com/fasterxml/jackson/core/exc/StreamReadException' (stack map, stack[0])

[2]

ftl.args.GcloudYmlTest > initializationError FAILED
    java.lang.NoClassDefFoundError: Could not initialize class ftl.test.util.FlankTestRunner

[bootstraponline]

Thank you for the detailed pull request and issue! I will take a look when I have some time.

[bootstraponline]

re: #632

Option 1 - Make the 10 second default significantly higher (2-5m)
Having an over-estimated default means that we err on the side of too many shards rather than too few. This effectively trades cost for stability. Enabling smart flank will mitigate this cost.

I prefer option 1 over introducing another flag. 10s was randomly chosen since it seemed to work well. Updating the default to something like 2m probably makes the most sense.

Flank already has a ton of config options. I am hesitant to add more unless we truly have a need.

[RainNapper]

re: #632

Option 1 - Make the 10 second default significantly higher (2-5m)
Having an over-estimated default means that we err on the side of too many shards rather than too few. This effectively trades cost for stability. Enabling smart flank will mitigate this cost.

I prefer option 1 over introducing another flag. 10s was randomly chosen since it seemed to work well. Updating the default to something like 2m probably makes the most sense.

Flank already has a ton of config options. I am hesitant to add more unless we truly have a need.

That sounds good to me. Simplifies the change too. I went with 2m.

[bootstraponline]

Thanks! That looks great.

I have to setup the bitrise jobs + CLA bot since we moved to the new Flank org. After that I'll merge this PR.

[bootstraponline]

Hi, sorry for the delay! We now have 2 full time resources working on Flank. 😄

I've merged the pull request.