webkit-gtk41 request for testing

[Greg Troxel <gdt%lexort.com@localhost>]

wiz@ asked me to revert changes to webkit-gtk41 as not adequately
justified, but I still think they are needed, so "more research is
needed" as the papers always say.

If you are up for test building, what's needed is:
  - pkgsrc-current, up to date packages, that you build
  - have PKG_DEVELOPER=yes
  - be able to look at `lib/libwebkit2gtk-4.1.so` in destdir to see how
    sqlite3 appears:
       - libsqlite3.so (right answer)
       - $prefix/lib/libsqlite3.so (odd; ldd works and PKG_DEVELOPER
         objects)
       - /path/to/work/dir/.buildlink3/lib/... (wrong; doesn't work, and
         PKG_DEVELOPER objects)
  - be able to build with make and ninja as cmake generators
  - this is complicated, so "understood the previous four points"

There are 4 ways to build on any given platform:
  - with or without the line forcing sqlite3 lib path, cross
  - with make or with ninja

So far:

  - NetBSD 10 amd64, make works with or without forcing, ninja fails
    with or without forcing (me)
  - NetBSD 10 amd64, make works (bulk on shadow)
  - NetBSD current, ninja works (wiz)


What would be interesting:

  - if make fails, with or without the forcing line, anywhere, in that
    the build is ok except the sqlite3 path in the lib is odd
  - any data about building with ninja, success or fail, ideally saying
    what happens with and without forcing

  - does bad sqlite3 path show up in any other file besides the main lib?

When freeze starts I will be updating my build domUs to pkgsrc-current
and can get data points there.