@PaulWessel, could you please clarify common option usage for a couple modules?

1. blockmean, blockmode, blockmedian include -g in the options macro, which is undocumented. My interpretation based on the docs and some tests is that this option does not provide any value for these modules. Could you confirm that this option can be removed?
2. Should gmtbinstats include -s? Currently -s is included and documented, but I cannot find a way to include empty bins in the output which would suggest that it should be excluded (similar to blockm* modules).

Also, is there a historical reason why -s is only available for output records, rather than input and/or output like -q, -d, -h, and -f?

@PaulWessel, could you please clarify common option usage for a couple modules?

1. blockmean, blockmode, blockmedian include -g in the options macro, which is undocumented. My interpretation based on the docs and some tests is that this option does not provide any value for these modules. Could you confirm that this option can be removed?

Yes, I cannot think of any reason -g is useful in this context.

2. Should gmtbinstats include -s? Currently -s is included and documented, but I cannot find a way to include empty bins in the output which would suggest that it should be excluded (similar to blockm* modules).

I agree with that too.

Also, is there a historical reason why -s is only available for output records, rather than input and/or output like -q, -d, -h, and -f?

Yes. It originated as a special -S option in grd2xyz to exclude output if a node was NaN. Then it migrated to a global common -s option in GMT 5 since there were other modules where not outputting a record if z = NaN made sense. There was no clear need for it to apply on input: If input records has NaNs GMT handles that very differently. NaNs are allowed except for in x-y coordinates I think. There is the GMT defaults IO_NAN_RECORDS [pass] that deals with how NaNs should be considered if found in input.