[m-rev.] diff: factor out common code in error_msg_inst.m

[Zoltan Somogyi]

On 2023-08-30 11:16 +10:00 AEST, "Peter Wang" <novalazy at gmail.com> wrote:
> I don't think there is any need to create two specialised procedures for
> each of these predicates; testing the flag at runtime will do.

There is no need when we use this code to generate error messages.
However, I am in the process of making a change to use this code more
when generating HLDS dumps, and *those* can take an annoyingly long time.
Since the cost is just a few duplicated mode declarations, I think making
HLDS dumps slightly faster is a good tradeoff.

Zoltan.