Add a reset() method to ProgressLoggerInterface.

[nh13]

I have two uses for this method:

1. I may create a ProgressLogger earlier, but want to set the startTime so it is right before the logger is being used. Yes, I can create the ProgressLogger at the last minute.
2. I may want to reuse a progress logger.

[codecov-io]

Codecov Report

Merging #1184 into master will increase coverage by 0.009%.
The diff coverage is 87.5%.

@@              Coverage Diff               @@
##             master     #1184       +/-   ##
==============================================
+ Coverage      68.4%   68.409%   +0.009%     
- Complexity     8014      8016        +2     
==============================================
  Files           541       542        +1     
  Lines         32690     32693        +3     
  Branches       5529      5529               
==============================================
+ Hits          22360     22365        +5     
+ Misses         8126      8125        -1     
+ Partials       2204      2203        -1

Impacted Files	Coverage Δ	Complexity Δ
.../htsjdk/samtools/util/ProgressLoggerInterface.java	0% <0%> (ø)	0 <0> (?)
...a/htsjdk/samtools/util/AbstractProgressLogger.java	42.593% <100%> (+2.208%)	7 <1> (+1)	⬆️
...samtools/util/AsyncBlockCompressedInputStream.java	76% <0%> (+4%)	13% <0%> (+1%)	⬆️

[tfenne]

LGTM. Let's give @jacarey and @lbergelson some time to weigh in, but if they don't object I'll merge this in a day or two.

[lbergelson]

This seems fine, it is a breaking API change which is gross, but I suspect the only downstream implementer of ProgressLogger is the GATK so it's probably not a huge deal in practice.

It feel like maybe this is a problem better solved by just creating it when it's needed, but maybe there's some good use for it that I'm not seeing.

[tfenne]

If you're worried about the breaking API change we could ask @nh13 to add a default no-op implementation of reset() in the interface.
I'll admit I would find this useful too on occasion. I have tools where I create 2-3 different loggers, one for each phase of work, and it would be be convenient to reuse a single one.

[pshapiro4broad]

You can prevent this being a breaking change by making reset() a default method in the interface. This is pretty much the use case default methods were designed for. Sorry, missed that @tfenne already covered this. Yes, please add a default method.

[pshapiro4broad]

In general, you want the javadoc on the baseclass or interface, so it's inherited by all subclassers and so a user can see the javadoc when dealing with the interface type.

[nh13]

Agreed, but I am going to pass on this change to be consistent with how ProgressLoggerInterface currently is written.

[lbergelson]

I agree with @pshapiro4broad but lets handle that separately.

[nh13]

I certainly can create the object right before it's used most of the time (I could use lazy in scala), but sometimes, there may be an initial delay from creation time to when the first record (and subsequent records) start processing that I don't want to affect the start time.

The alternative I was considering was having a way to configure ProgressLogger to set the startTime upon recording the first record.

[lbergelson]

@nh13 This seems like a fine solution, setting it when you record the first record is usually alright, but it has a similar issue to setting it at construction time if for some reason the first record is unusually slow or if you only have a few long operations being recorded. 👍 to this change with the default method.