Check number of repeats for code examples

[JostMigenda]

We currently use repeats = 1000, which can be extremely slow (half an hour or so) on a few examples (e.g. linear_search_list and uniqueList).

We should reduce those to a few min at most. (Personally, I wouldn’t make them too quick, though—that could easily give the impression that “well, it’s technically slower; but it doesn’t matter in practice because they’re both instant”. Making fast things feel fast and slow things feel slow can help drive home the point more effectively.)

Based on comment by @Robadob in #14 (comment)

[Robadob]

In general, the main value is putting across that one can be much slower than the other. Repeats is just there to get a stable time. In most cases, it's poor scaling, such that increase N an order of magnitude and the speedup will get even higher. (As such, there's little harm if you're using the bad approach in a limited way but if profiler says its a large part of your runtime, that's why).

An initial idea had been to go more complicated than fixed benchmarks, and produce graphs to demonstrate the scaling. However, this would effectively require writing a package graphit or equivalent to allow this to be achieved with minimal code. Furthermore, some of the scaling (such as set vs list for handling uniques), depends on the data being processed so would require multiple plots to show this. So it quickly felt non-viable.