run_parallel uses pmap and handle worker ids correctly

[lassepe]

• use progress_pmap from ProgressMeters.jl
• introduc pool argument to specify the worker pool
• fix the bug of wrong worker id's by iterating over pool instead of 1:nprocs() (not all ID's must be occupied, e.g. if workers were killed).
• update warning to match the new behavior

[coveralls]

Pull Request Test Coverage Report for Build 52

• 9 of 9 (100.0%) changed or added relevant lines in 1 file are covered.
• 7 unchanged lines in 1 file lost coverage.
• Overall coverage decreased (-1.0%) to 82.891%

---

Files with Coverage Reduction	New Missed Lines	%
src/stepthrough.jl	7	66.23%

Totals
Change from base Build 43:	-1.0%
Covered Lines:	281
Relevant Lines:	339

---

💛 - Coveralls

[zsunberg]

Looks good, except progress isn't actually fed into pmap_progress. After thinking about it last night, I actually like the idea of having progress and show_progress as separate arguments. Can you make the change to that?

Thanks! Thanks for writing good code

[lassepe]

Looks good, except progress isn't actually fed into pmap_progress. After thinking about it last night, I actually like the idea of having progress and show_progress as separate arguments. Can you make the change to that?

Thanks! Thanks for writing good code

Okay, will do. Note, that implementing it with show_progress will change behavior. In the current master version, if one passes progress=nothing, no progress bar is shown. (this behavior is also used in the tests).

[zsunberg]

Great, thanks @lassepe . I just added a deprecation warning for progress=nothing in case it is being used anywhere

[zsunberg]

If the tests pass, it looks good to me