Re: speed up unicode decomposition and recomposition

Attached v3 addressing review points below:

On Thu, Oct 15, 2020 at 11:32 PM Michael Paquier <michael(at)paquier(dot)xyz>
wrote:

> + # Then the second
> + return -1 if $a2 < $b2;
> + return 1 if $a2 > $b2;
> Should say "second code point" here?
>

Done. Also changed the tiebreaker to the composed codepoint. Beforehand, it
was the index into DecompMain[], which is only equivalent if the list is in
order (it is but we don't want correctness to depend on that), and not very
clear.

> + hashkey = pg_hton64(((uint64) start << 32) | (uint64) code);
> + h = recompinfo.hash(&hashkey);
> This choice should be documented, and most likely we should have
> comments on the perl and C sides to keep track of the relationship
> between the two.
>

Done.

> <separate headers>

Done.

Other cosmetic changes:
- format recomp array comments like /* U+0045+032D -> U+1E18 */
- make sure to comment #endif's that are far from the #if
- small whitespace fixes

Note: for the new header I simply adapted from unicode_norm_table.h the
choice of "There is deliberately not an #ifndef PG_UNICODE_NORM_TABLE_H
here", although I must confess I'm not sure what the purpose of that is, in
this case.

--
John Naylor
EnterpriseDB: http://www.enterprisedb.com
The Enterprise PostgreSQL Company