kernel: add 'static' keyword where possible

[fingolfin]

This PR is somewhat dual to PR #3183 resp. #3202: it adds static to as many functions and variables as possible. This helps to find dead code (see e.g. PR #3203), may help compilers to do better optimizations, and also is a step towards defining a larger "libgap" API, by "hiding" away stuff that we do not explicitly want to export (this makes it harder to "cheat" and call those functions anyway; now people who want to do that hopefully will instead talk to us to get us to export them, giving us a chance to define a proper API).

The second commit is optional and clang-formats the first commit, based on the idea that if we touch those lines anyway, we might as well reformat them, too.

[codecov]

Codecov Report

❗ No coverage uploaded for pull request base (master@e868e29). Click here to learn what that means.
The diff coverage is 86.24%.

@@            Coverage Diff            @@
##             master    #3204   +/-   ##
=========================================
  Coverage          ?   84.76%           
=========================================
  Files             ?      687           
  Lines             ?   335795           
  Branches          ?        0           
=========================================
  Hits              ?   284634           
  Misses            ?    51161           
  Partials          ?        0

Impacted Files	Coverage Δ
src/vector.c	92.73% <ø> (ø)
src/vars.c	93.3% <ø> (ø)
src/vecgf2.c	73.3% <ø> (ø)
src/weakptr.c	98.24% <ø> (ø)
src/vecffe.c	71.02% <ø> (ø)
src/vec8bit.c	83.91% <ø> (ø)
src/objcftl.c	23.86% <0%> (ø)
src/integer.c	98.43% <100%> (ø)
src/bool.c	98.94% <100%> (ø)
src/code.c	96.28% <100%> (ø)
... and 52 more

[coveralls]

Coverage decreased (-0.0003%) to 84.643% when pulling e8dee1e on fingolfin:mh/static into e868e29 on gap-system:master.

[ChrisJefferson]

Did you check this didn't break compiling / loading any packages?

[fingolfin]

@ChrisJefferson the testpackages test on Travis verifies that all distributed packages still compile and load. This testing really works: it even found one issue in an earlier revision of this PR: the cvec package unfortunately has extern unsigned long PolsFF[]; so I had to drop static from PolsFF -- at the same time, this is a nice example for exactly the kind of thing I'd like to prevent in the future: I'll work on fixing cvec to not require this (possibly by adding a suitable new API to the GAP kernel). Actually, I was already aware of this before, see gap-packages/cvec#17. I'll try to get it addressed in time for GAP 4.11