addQuitProc now works with closures, and c, js(node/browser) backend; fix some bugs in testament

[timotheecour]

• fixes all issues in Documentation nonexistent quitprocs module #14331 (comment)
  (in particular ordering issue)
• it's moved quitprocs under std (as for all new modules)
• avoid the need for a different overload
• remove the C limitation of 30 atExit calls; (it should now be more efficient although untested)
• make it work for -d:js (node and browser)
• add test

testament fixes

• testament/testament r tfoo now correctly adds target-specific default options (for now just -d:nodejs for js), just like it does when running without r (eg the js test suite runs with -d:nodejs but it didn't add -d:nodejs for testament r); this was needed so that this works: nim r testament/testament r tests/stdlib/tquitprocs.nim (+ others)
• minor refactors to avoid some code duplication around areas I had to change, and fixes formatting (missing space) when extra options are passed: tests/stdlib/tquitprocs.nim JS--lib:lib => tests/stdlib/tquitprocs.nim JS --lib:lib

caveat

caveat is pre-existing from when addQuitClosure was added, and seems consistent with the preexisting doc comment: In case of an unhandled exception the exit handlers should not be called explicitly, although I think behavior should be changed (in future PR)

if a registered quitProc raises during teardown, subsequent procs won't be called; one option (in subsequent PR) would be to enhance this by wrapping each call into a try ... except and keeping track of at least one raised; and then call quit(0) or quit(1) depending on that

CI failures unrelated

• tests/stdlib/tfdleak.nim is flaky timotheecour/Nim#190
• tests/vm/tslow_tables.nim is flaky (at least on windows) timotheecour/Nim#191

TODO for future PR's

• revisit behavior when a registered proc raises
• add a check in kochdocs that new modules are added only under std/; there's already all logic in place for globbing source files + checking against whitelists
• add vmops for addQuitProc; I've been needing this a few times, eg in some patch I've added some counters (timers + stats etc) around key compiler procs and the timers should be shown at exit

[Araq]

Please don't merge the two separate mechanisms into one.

[zah]

It would be quite useful if the same treatment is applied to procs such as setControlCHook.

[timotheecour]

@zah

It would be quite useful if the same treatment is applied to procs such as setControlCHook.

this is left as future work

@Araq
PTAL, all comments addressed.
order is sane so long you use addExitProc (whether using a regular proc or closure);
I'm deprecating addQuitProc because it shouldn't be in system in the first place, and because you can't guarantee a sane ordering (last in first out) if you mix regular vs closure procs; however existing code won't break.