py: use subprocess to read with timeout

[milahu]

@mkoeppe please check before we merge this. ./py/test/test.sh looks good

previously i wanted to avoid "subprocess to read with timeout"
but starting a new python interpreter from python uses almost no extra memory
because the python executable is shared memory
so this uses even less memory than calling the dd executable (5MB RSS)

same python interpreter instance running multiple scripts simultaneously

when you run two or more copies of the interpreter, the second and subsequent will share most of the memory (at least for executable code) with the first, so resource usage will not rise (anywhere close to) linearly as you run more instances.

pros:

• avoid messing with signals. no more need for cysignals
• no more need for os.dup(self._fdRead)
  blocking os.read is guarded by timeout of subprocess
• portable: should also work on windows
  → fix py: fix read with timeout on windows #10
• the same concept also works in javascript
  → js: fix read with timeout on windows #6

cons:

• may be slower than the previous solution, but this is no hot code

_parse_exception is needed to handle unexpected errors from the python subprocess

[mkoeppe]

It would be easier for the use in Sage if we could accept (and ignore) use_cysignals for a while

[mkoeppe]

see sagemath/sage#38116 (comment)

[milahu]

just help yourself : )
i dont use this, just sharing my drafts