Convert SQL to Conjunctive Queries

The current scripts do two things:

1. "Normalizing" a SELECT-FROM-WHERE query, by striping away everything that is not a join predicate (t1.col1 = t2.col2) from the WHERE clause.

2. Translate the normalized query to CQ (a.k.a. tensor algebra), dropping any column that does not participate in any join. If we kept all columns the CQ would be way too large, and we would generate extremely deep loop nests in TACO.

3. Compile with cargo build --release (install cargo here)

4. Enter pipenv with pipenv shell (install pipenv here)

5. Run with cargo run 9b.sql, where 9b.sql is a file containing 1 SQL query.

6. Use fmtsql.py to format the SQL output; run sed -E 's/(\_|\.)//g' 9b.cq with 9b.cq containing the CQ output to strip . and _ away from the CQ.