Misc performance improvements for CRAM reader

[athos]

This PR tweaks the performance of the CRAM reader by the following changes:

• 16c4a00: Use lookup tables to look up rANS symbols, instead of doing binary search
• 2221b5e: Replace maps and vectors used in the rANS decoder with arrays
• 2dc6ae7: Coerce decoded quals to make = more efficient
• 154426f: Replace + with bit-or

Overall, these changes make the CRAM reader roughly 4x faster than before on my environment:

(with-open [r (reader "DRR002191.chr1.sorted.cram" {:reference ".cavia/hg38.2bit"})]
  (time (count (take 1000000 (read-alignments r)))))

;; before change
"Elapsed time: 22955.078416 msecs"

;; after change
"Elapsed time: 5079.17675 msecs"

Here are the profiling results before and after the change:

before change	after change

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 88.67%. Comparing base (43df52d) to head (154426f).

Additional details and impacted files

@@           Coverage Diff           @@
##           master     #304   +/-   ##
=======================================
  Coverage   88.67%   88.67%           
=======================================
  Files          93       93           
  Lines        8098     8100    +2     
  Branches      505      505           
=======================================
+ Hits         7181     7183    +2     
  Misses        412      412           
  Partials      505      505           

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[alumi]

Thank you for the Pull Request!
With the speed improvements you implemented, I believe it's becoming more practical to use. Great job! 👍

[r6eve]

LGTM! Thank you for the PR.