Add intra-run multithreading to SearchCollection

[lintool]

Context: SearchCollection currently has the ability to do parameter sweeps, with different configurations in parallel using a thread pool. However, it currently has no ability to run different queries in a single configuration in parallel. This means that some sparse models take a long time to run.

This PR adds a threadpool within an individual configuration to run queries in parallel.

[codecov-commenter]

Codecov Report

Merging #1678 (5bad21d) into master (f7970fb) will decrease coverage by 0.13%.
The diff coverage is 51.54%.

@@             Coverage Diff              @@
##             master    #1678      +/-   ##
============================================
- Coverage     57.36%   57.23%   -0.14%     
  Complexity     1029     1029              
============================================
  Files           172      172              
  Lines          9618     9668      +50     
  Branches       1324     1336      +12     
============================================
+ Hits           5517     5533      +16     
- Misses         3637     3666      +29     
- Partials        464      469       +5     

Impacted Files	Coverage Δ
...rc/main/java/io/anserini/search/SearchMsmarco.java	0.00% <0.00%> (ø)
...main/java/io/anserini/search/SearchCollection.java	43.92% <51.64%> (-1.28%)	⬇️
...c/main/java/io/anserini/index/IndexCollection.java	45.01% <100.00%> (ø)
src/main/java/io/anserini/search/SearchArgs.java	100.00% <100.00%> (ø)

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update f7970fb...5bad21d. Read the comment docs.

[lintool]

All regressions successfully passed with the multi-threaded implementation.