HDDS-3762. Intermittent failure in TestDeleteWithSlowFollower

[adoroszlai]

What changes were proposed in this pull request?

Intermittent failure in testDeleteKeyWithSlowFollower seems to be caused by:

• DeleteBlocksCommandHandler increments invocationCount near the beginning of handle(), and only updates deleteTransactionId later
• TestDeleteWithSlowFollower waits for invocationCount >= 1, then asserts deleteTransactionId also increased

The test is fixed by changing the order in the handler: only increment invocationCount at the end, when deleteTransactionId is already updated.

I think invocationCount should be updated together with totalTime to provide (slightly more) correct average run time (getAverageRunTime).

https://issues.apache.org/jira/browse/HDDS-3762

How was this patch tested?

Passed 50x:
https://github.com/adoroszlai/hadoop-ozone/runs/1059570465#step:5:4

Regular CI:
https://github.com/adoroszlai/hadoop-ozone/runs/1059560880

[bshashikant]

Thanks @adoroszlai for the contribution.

[adoroszlai]

Thanks @bshashikant for reviewing and committing it.