Add query preprocessing script for embedding generation

[J-Dymond]

This pull request introduces a preprocessing utility for generating query embeddings from IR datasets:

New Script: preprocess_queries.py:

### Usage: python arc-exps/preprocess_queries.py --json_config_path [config path].json

The config json file should contain:

{
    "embed_model": "msmarco-distilbert-base-tas-b",
    "data": {
        "dataset": "msmarco",
        "query_set": "msmarco-document/trec-dl-2019",]
    },
}

Outputs are saved with the following file structure:

processed_queries/
├── {dataset}/
│   ├── {query_set}/
│   │   └── {model_name}.csv

The file output is a csv file with 3 fields, query-id, query text, and the embedding vector.

Slurm Scripts:

Also added are slurm scripts for running jobs on computing clusters with slurm job scheduling.

scripts/slurm/pre-process-embeddings.sh can be used to run preprocess_queries.py and takes the path to the config as an argument.

[eddableheath]

All LGTM!