Re: Upper limit arguments of pg_logical_slot_xxx_changes functionsaccept invalid values

The following review has been posted through the commitfest application:
make installcheck-world:  tested, passed
Implements feature:       tested, passed
Spec compliant:           not tested
Documentation:            not tested

The error messages are nice, and I think will be a helpful feature. This patch does add them correctly and also tests
them.

A few things I noticed, in the code, that don't really interfere with functionality are below:
In the `if (PG_ARGISNULL(1))` section:
I was wondering if creating an alias called something like InfinateInvalidXLogRecPtr might be better than adding a
commentsaying infinite.
 

In the `if (PG_ARGISNULL(2))` section:
I think '0' is being used as a magic number here and could probably have another #define (linked to a place to put it
below).Also, it seems a little weird to take NULL and then just set upto_nchanges to '0' while disallowing '0' to be
passedin as the argument. Maybe just allowing '0' as an input would be ok and useful to some people. I don't believe I
wouldhave noticed that if 'upto_nchanges = UnlimitedNchanges` instead of 'upto_nchanges = 0'.
 

 https://doxygen.postgresql.org/xlogdefs_8h_source.html#l00028