Build out cosDPR-distil regressions for TREC 2019 and TREC 2020 for Anserini

[lintool]

Here is a concrete task. If you look at our Anserini regressions, under "MS MARCO V1 Passage Regressions", we have missing entries for TREC 2019 and TREC 2020.

The concrete task is to build these regressions.

We need to encode the queries and then convert them into Anserini's JSON format. @MXueguang might actually have them encoded already (in numpy?)... in which case we just have to convert them over.

Warmup tasks:

(1) Reproduce cosDPR-distil on MS MARCO: https://github.com/castorini/anserini/blob/master/docs/regressions/regressions-msmarco-passage-cos-dpr-distil.md - make sure you can get it running on student linux env.
(2) To understand the context of what you're doing, read: https://cs.uwaterloo.ca/~jimmylin/publications/Ma_etal_CIKM2023.pdf

This is related to #3 - @pratyushpal and @mchlp you might be interested.

[pratyushpal]

@lintool I'll work on it!

[pratyushpal]

This task was solved in this PR : castorini/anserini#2204

[lintool]

This is done, closing.