Using a fraction of read length for the maximum allowed mismatch

[mengzhou]

Currently the -m option is an integer indicating the number of mismatch allowed per read. But in practice it is better for this number to be automatically determined based on input read length. In rmap, this option can be a float number, e.g. 0.1, to indicate the fraction. Therefore by default the maximum mismatch per map is read_length * 0.1.

For walt it is necessary to calculate such number per read during mapping, because it is possible that the input reads are processed after adapter trimming, and have different lengths.

[acgtun]

This is a good idea.