cad/gmsh: upgrade to 2.13.2, fix knob issue, other various changes

[jrm]

Summary

• Upgrade to version 2.13.2
• Update MASTER_SITES to only include new gmsh site
• Remove no-longer-needed definitions of DISTFILES, EXTRACT_ONLY, and WRKSRC
• The licensing situation is simpler now: Gmsh itself is released under GPLv3+ [1], TetGen under AGPLv3+ [2], and METIS under an APACHE2 license [3] For now, TAUCS is the only component released under a non-standard license [4]
• Fix misnamed KNOB in KNOB_DESC=, so make config shows descriptions
• Fix errors/warnings reported by portlint and poudriere testport
  • Add missing entries to LIB_DEPENDS=
  • Add blaslapack to USES=
  • Change USE_GL=yes to USE_GL=gl glu
  • Add USE_XORG=x11 xext xfixes xft
  • Redo patches with make makepatch
• Remove MAKE_JOBS_UNSAFE=yes (no build errors)
• Match upstream by turning on TAUCS by default
• Modernize option handling
• Add patch to fix linking error when the TAUCS knob is turned on
• Mark broken on FreeBSD 12 i386 due to clang bug: (https://llvm.org/bugs/show_bug.cgi?id=21903)
• Remove unnecessary TAUCS_LIB_DEPENDS=libtaucs.so:math/taucs to prevent problems with french/aster
  • cad/gmsh includes a bundled taucs, so TAUCS_LIB_DEPENDS=libtaucs.so:math/taucs can be removed
  • math/taucs pulls in math/metis4, but french/aster, which depends on cad/gmsh ad/gmsh, pulls in conflicting math/taucs-edf
• Update pkg-plist

porlint -C: OK
testport: OK (poudriere: 9.3-RELEASE, i386, default options)
testport: OK (poudriere: 9.3-RELEASE, amd64, default options)
testport: OK (poudriere: 10.3-RELEASE, i386, default options)
testport: OK (poudriere: 10.3-RELEASE, amd64, default options)
testport: OK (poudriere: 11.0-RC2, i386, default options)
testport: OK (poudriere: 11.0-RC2, amd64, default options)

http://pkg.awarnach.mathstat.dal.ca/data/latest-per-pkg/gmsh/2.13.2/

[1] http://gmsh.info/#Licensing
[2] http://wias-berlin.de/software/tetgen/1.5/FAQ-license.html
[3] http://glaros.dtc.umn.edu/gkhome/metis/metis/faq?q=metis/metis/faq#cite
[4] http://www.tau.ac.il/~stoledo/taucs/

Test Plan

While cad/gmsh builds fine for all the architectures/versions listed above, it can fail on live systems when ceratin (yet-to-be-determined) packages are installed. I hope to investigate this before the next patch.

Before committing, sort and align text.

[mat]

Please, remove all the whitespace changes you added it makes it impossible to read the diff

[jrm]

This one should be easier to see what's going on.

[mat]

Makefile

32 ↗

(On Diff #19945)

Are you sure the issue for which it was marked unsafe has been fixed ?

80–82 ↗

(On Diff #19945)

This should go above the implies, so that it is readily readable.

[jrm]

Makefile

32 ↗

(On Diff #19945)

I just tested the builds. The commit message for the commit that added the line doesn't have any information. Is there anything else I can test?

[jrm]

Move comment describing why TAUCS_IMPLIES=METIS is necessary inline.

[jrm]

Update diff after r422711 and r422828 were committed

[swills]

LGTM, Approved

[jrm]

Incorporate suggestions from AMDmi3

• Remove options for bundled software
• Update default options
• Use OPT_CMAKE_BOOL

[swills]

LGTM, Approved