Always print ReferenceCounter info for leaks

[roystgnr]

This info is kind of trivial (and so tempting to suppress via the command-line option) when there's not a leak, but in the event there is a leak, we shouldn't be hiding any details.

See the discussion in idaholab/moose#32877

[roystgnr]

If @jwpeterson agrees, I'd even be happy to go as far as defaulting to disable_print_counter_info() when there's no leak (with an optional --enable-refcount-printing argument to override, instead of the current --disable- argument). The reference counter info is occasionally interesting for profiling and it's often interesting for debugging leaks but 99.9% of the time it's just clutter.

[moosebuild]

Job Coverage, step Generate coverage on 78b0900 wanted to post the following:

Coverage

	5bfc97	#4455 78b090
	Total	Total	+/-	New
Rate	65.50%	65.49%	-0.00%	100.00%
Hits	78295	78292	-3	5
Misses	41245	41252	+7	0

Diff coverage report

Full coverage report

This comment will be updated on new commits.

[loganharbour]

Works for me. I think the quiet behavior being the default is a good idea

[jwpeterson]

If @jwpeterson agrees, I'd even be happy to go as far as defaulting to disable_print_counter_info() when there's no leak

I can't remember the details, but I think there was a situation (maybe in MOOSE?) where "Memory leak detected!" was being printed at the end of every simulation, but it either wasn't "real" and/or there was nothing we could actually do about it. So that was why we added the ability to disable the leak message entirely.

I searched for the first instance of disable_print_counter_info() and found 0cc5cec (from 2010) and it refers to "Paul", so maybe it was GRINS-related? Anyway, I'm fine with the update, but I guess we'll need to see if there's more false positives once it's turned on all the time.

[loganharbour]

Should we print this to err instead now that we're only printing in the case of leaks? as the next message is to err too

[roystgnr]

That's a really good idea. A big leak on a big parallel run will spew a ton of output now, but that's much better than just getting the error message without any hint of the cause if the leak doesn't happen to trigger on rank 0.