[CI] now enables NIM_COMPILE_TO_CPP=true to run without allow_failures

[timotheecour]

• [CI] now enables NIM_COMPILE_TO_CPP=true to run without allow_failures

• adds workaround for nim cpp -r tests/exception/t9657 hangs #10343

• fixes bug caused by csources being built assuming raise_id even though stdlib was updated in meantime to use raiseId ; this workaround can be removed after the next release of csources (but nothing will break if it's not removed)

• requires these other PR's the be merged first: EDIT: DONE!

  • fix sdl_test test that failed in CI cpp mode #10314 sdl_test (MERGED)
  • fix twrong_refcounts in nim cpp mode #10313 twrong_refcounts (MERGED)
  • fix #10305 nim cpp is now nan-correct at CT #10310 (MERGED)
    • tests/float/tfloatnan.nim
    • lib/pure/math.nim
    • tests/float/tfloatmod.nim

[Araq]

I'm taking over, stay tuned.

[timotheecour]

rsiseId gives error when building from csources, as i noted above, with cpp mode

[timotheecour]

superseded by:

• 8947779 (for the test disabling)
• 2039dad (but that's slower than what I suggested above, as it adds another iteration in bootstrap, which affects CI which is already resource constrained, see https://travis-ci.org/nim-lang/Nim/jobs/480560543)

[timotheecour]

let me try something, please keep open for now

[timotheecour]

@Araq PTAL

• I fixed that typo above(raise_id vs raiseId)
• the latest fix I'm doing in this PR is better than 2039dad because it requires 1 fewer iteration of building nim, which is critical for saving on CI time (as well as on user's machines); it'll also automatically become a noop once csources get updated (without having to change anything)
• also fixed another blocker for CPP
• most importantly, NIM_COMPILE_TO_CPP=true is green for the 1st time!