Remove OML_INCR_COUNTER mention of IFS routine

[wdeconinck]

[Originally authored by @flpvana]

This pull request improves the OML_INCR_COUNTER subroutine in the OML_MOD module, focusing on better error handling and documentation cleanup. The main change is the addition of an optional error output argument to allow external error handling, which makes the routine safer and more flexible. There are also minor corrections to argument intent and documentation.

Error handling improvements:

• Added an optional KERR argument to OML_INCR_COUNTER to allow the caller to handle errors externally, instead of always aborting the program. The subroutine now sets KERR to indicate an error if K exceeds KMAX, and only aborts if KERR is not provided.

Documentation and intent corrections:

• Updated the documentation to remove references to an IFS routine in OML_INCR_COUNTER and documented the new KERR argument.
• Corrected the INTENT of the MYEVENT argument in OML_WAIT_EVENT from INOUT to IN for clarity and correctness.

[marsdeno]

Looks fine, although we are still presupposing names of lock-related variables at call site.

[flpvana]

The best version so far. I would not mind the two different names of compared variables. Both are correct, the first one for users, the other one for bug hunters finding the error message.

[wdeconinck]

The routine has actually nothing to do with OML internally it seems; so in principle this could all be handled at call site directly... Fine with this though.