Fix for Gctoolkit #372 - YoungDetails event not published for GC log collected with -Xlog:gc (no *)

[jlittle-ptc]

Approach:

Both files that are generated using -Xlog:gc and -Xlog:gc* use the same general format for the log line of interest:
-Xlog:gc: [6.837s][info][gc] GC(3) Pause Young (Normal) (G1 Evacuation Pause) 79M->27M(1024M) 2.929ms
-Xlog:gc*: [1.361s][info ][gc ] GC(0) Pause Young (Normal) (G1 Evacuation Pause) 18874M->9037M(81920M) 23.698ms

The main difference is that for -Xlog:gc, this is the only line that is printed for an event, whereas -Xlog:gc* prints a CPU time line right afterward, which GCToolkit currently uses as a publish trigger:
[1.361s][info ][gc,cpu ] GC(0) User=0.95s Sys=0.09s Real=0.02s

To address this issue, I've added a Diarizer flag PRINT_CPU_TIMES that is populated when extracting decorators. If this flag is unset/unsure, the UnifiedG1GCParser.youngDetails() method will populate remaining fields (GC Type - Young, GC Cause extracted from trace 4 (offset -2), StartTime as reported by getClock()) and attempt to publish the event. I have left in a null check on GCType in the publish condition as a failsafe, but in theory it could be removed.

I think this should be reasonably solid, is a relatively minor change, and it will keep events being published in order.

Test Updates
Not much done here, other than adding a couple of updates to existing test cases. I can provide a sample file, and can update some additional suites, though I wanted to confirm how I can go about doing that. Is it just a matter of a PR to gctoolkit-testdata with the new file first?

[kcpeppe]

This looks good, thank you! There is additional work that needs to be done but I'll create a new issue to describe that. As for test data, yes, you will need to add the log files to gctoolkit-testdata. Once that is in, I'll trigger a release.

[kcpeppe]

Needs a follow on issue to fix diarization tests for the Generational, Z, and Shenandoah collectors.