Improved mapping results for a multi-part index - merging

[hasindu2008]

The following features were added (for uni-segment reads only). The additions were done such that minimal changes are done on existing code.

1. --multi-prefix option to specify a temporary file prefix
2. Dumping the internal state of each mapping to temporary files
3. Merging the intermediate values to generate the final output (merged headers, fixed mapq, fixed flags and the printing of the correct number of the secondary mappings/unaligned reads)

Tested using two datasets 33k reads and 3.9M synthetic long reads. Tested to confirm that the changes do not affect the existing result (see the attached evaluate_accuracy.zip)

evaluate_accuracy.zip

[lh3]

Thanks so much for this. I will review and merge later, definitely getting this ready for the next release 2.12.

[lh3]

Learning from your code, I have reimplemented the support of multi-part index at the split branch. The new option is --split-prefix, which has the same behavior as your --multi-prefix. If you have time, you may test it. I will merge this branch to "master" once I confirm it is stable.

[lh3]

I have merged the split branch to master. I will close this PR. If you find bugs in the new feature, please create an issue. Thank you very much for this PR. It has been very helpful.