Adds parallel map-reduce recipe

[j-haj]

fixes #329

• Adds example in concurrency.md
• Adds links to intro.md

[budziq]

Very cool. Just few nits below

[budziq]

Very cool. Just few nits below

[budziq]

I'd suggest mapping first and then using filter(gt_30) instead of closure in closure

[budziq]

I'd move this comment to the textual description above

[budziq]

I'd suggest using std::f32::EPSILON instead of 1e-8

[budziq]

I'd probably move casts to above and store the variables as f32 but it's a matter of taste.

[j-haj]

I run into a type inference issue with the alt_sum_30 variable if I moved as f32 to the line where alt_sum_30 is initialized. So far the easiest way to get rid of that error is to make alt_sum_30 a u32 and then convert to f32 later. Any suggestions?

[budziq]

right sum is a template so type inference gets wonky here.

 let alt_sum_30: f32 = v.par_iter()
        .filter(|x| gt_30(x.age))
        .map(|x| x.age as f32)
        .sum();

or

    let alt_sum_30 = v.par_iter()
        .filter(|x| gt_30(x.age))
        .map(|x| x.age as f32)
        .sum::<f32>();

or the cleanest solution, leaving is as it was, summed into u32 and casting later on ;)

[budziq]

the &ref and ref are not needed anywhere in this example

[j-haj]

Thanks for all the comments! I have pushed an updated version

[budziq]

Nicely done!