standardise number of elements sorted

[roarin-roran]

number of elements to sort should be a control parameter, and should be standardised.

between 10k and 10 million elements is usually good (should be noted in the lab book, in doccumentation, and printed)

default: 10k, should be a global parameter

[roarin-roran]

seb, when asked "how many elements should we sort?:"

depends on the application.

for running time experiments shouldn't be too small so that we don't (only) get in-cache performance. 1-10M elements should do.

less critical for combinatorial cost measures, but have to avoid cutoff parameters such as minrunlen 10-100k often enough.

[roarin-roran]

10k is default currently: problem is that this is not strictly enforced in inputs. test inputs to ensure this is actually used

[roarin-roran]

complete - one element found which didn't do this, further tests will be completed as part of finalising the catalogue