Log message:
fcitx5: Fix build failure with fmtlib > 12.0.0 (pkgsrc version: 12.1.0)

Don't use optional code hunk that adds timestamps to logs for newer fmtlib.
The recommended migration path is for std::localtime, but this seemed like
a patch that I'm less likely to get wrong, as someone who is not experienced
in C++.

Log message:
*: recursive bump for pcre2

Running an old binary against the new pcre doesn't work:
/usr/pkg/lib/libpcre2-8.so.0: version PCRE2_10.47 required by \ 
/usr/pkg/lib/libglib-2.0.so.0 not defined

Log message:
*: recursive bump for fmtlib 12

Log message:
*: recursive bump for tiff growing lerc dependency

Log message:
*: fix build with cmake 4

Log message:
*: recursive bump for jpeg -> libjpeg-turbo switch

Log message:
fcitx5: do not build tests

Fixes build on NetBSD/current.

Log message:
devel/extra-cmake-modules: Use buildlink3.mk, not TOOL_DEPENDS.

This one is borderline: should it be BUILD_DEPENDS or TOOL_DEPENDS?
It doesn't appear to vary depending on the build environment, so it
doesn't meaningfully describe the cross-builder vs the target system.
It is used only at build-time, but it doesn't have any native
executables -- they're just source code cmake scripts.

For now, it is convenient to use a buildlink3.mk file because this
enables us to explicitly expose only files under share/ECM/ -- and
not other random files installed in the prefix -- via buildlink3 to
keep dependent package builds clean.  Eventually we should have the
same mechanism for TOOL_DEPENDS but that day has not yet come.

This change creates a new use of BUILDLINK_DEPMETHOD build for
extra-cmake-modules, which is a priori risky.  Fortunately,
extra-cmake-modules has no libraries, so this is unlikely to cause
trouble.