Regenerated jcstress playlist for jcstress and pushed its generator (a…

…doptium#5317)

* Regenerated jcstress palylist for jcstress and pushed its generator

The initial set of tests is obtained by `jcstress.jar -l`
The generator is:
  mixing targets with  to few targets
  ensuring no test is run twice
  targets are as short as possible where still fully descriptive
  regexes for -t runtime are fully quified

Similarly as there is exclude list to not shorten groups where
missleading preffix would remain, can be added include list, which would
connect remaining tests with 1 or two targets to artificial group. But
it was jsut 23 tests in time of this commit

* Small spell check over generate.java

* in initial spirit, rewrote Generator to be more fair

* more strict checksum
* merging of groups go until it really have nothiong to do
* except natural groups, also artificial are created
* some control env variables like verbose, limit and jsut_regex to check
  times
* rearranged code

* Updated generator to contain self check

* it can launch the jcstres on all targets it generated, and meassure
  times
* on my machine, in single core mode, it is 20+-1minutes per target (in all
  cores mode (8 cores) it is 13 hours per target (not even one
finished(yet)
* fixed selector generator to contains or always

* Made cores adjsutable also for playlist.xml

* Set default number of cores for self-test to two

* Fixed selector.

Now spliting works as expected. On two cores:
splitting by 100, 54 groups 4hours each
splitting by 2000 7 groups 1.5 day each
no splitting  11.5 days total

Oscialtion to be calculated

* Lowered default grouping limit to 10 and updated main comment

* Returned limit to 100. Much better naming.

* Enhanced generator to calculcate ETA without actually running the suite

* Finished Generator to split fairly by tests and to print real time-based estimations

* Not spliting test and runtime cores. One CORES for both.

Use dalso in -l, but -l do nothonour -c

* Split test sum to informative arbiters and actors

* Disabled non FQN based regexes implicitly

* Instead of removing FQ from FQN, reusing longest known preffix

This is saving quite a few from selector, seesm to work fine and seems
to not have overlaps

* Checking hashmap type

* Minor correction of appending to details file

* Improved generator to work with -tb; regenerated  for some 3 hours (1test 0.9 s)

* Added warning about imprecise -tb

https://bugs.openjdk.org/browse/CODETOOLS-7903750

* Added MAX_NATURAL_ITERATIONS  limiting maximal ccompression on natural groups

some refgactoring senaked in...

* generated playlist by latest agrrement and added readme.md

adoptium#5317 (comment)

* Small reformatting

* Mentioned the checksum

* Added warning about shell bnoundaries

* spell check

* Mentioned command

* moved disable out of xml comment

* Removed duplicated $(APPLICATION_OPTIONS)  usage

* Renamed main target to jcstress.all

* Fixed semicolons and brackets

* Fixed env vars handling

* Removed unused export

* Added todo about jcstress.console.printIntervalMs

* Disbaled also main target

* Set default cores to 2

Despite usptream disregards this, it is currently the only way to cut
runtime to at least somehow reasonable duration

* added ability to validate

* Regeenrated playlist

* Fixed buggy stderr outpu

* Subtargets now also honour timebudget

* jcstrress' loging  slowered to once per minute (from 15s)

* IMproved handling of timebudget so it can no longer be confused