Three src/pl tests have comments about how they fail in EUC_* encodings other
than EUC_JP_2004. I think this predates psql \if and \gset meta-commands
making test skips less onerous, so let's skip using those, as attached.
Why now? Commit c67bef3 added euc_kr.sql to exercise some code specific to
non-UTF8 multibyte encodings. I want it to be possible to exercise that code
with settings that also pass check-world as a whole. The alternative was to
use EUC_JP_2004 or maybe MULE_INTERNAL to exercise that code. EUC_KR is
closer to present-day relevance than MULE_INTERNAL or EUC_JP_2004, which don't
have glibc locales.