Improve query workload identification

[ian-whitestone]

There's a number of things we could do to improve our workload detection. Right now we just strip out comments and then group by query text, but we can offer other improvements like:

• Remove database + schema names from all tables in a query (account for users running the same workload on cloned datasets).
• For CTAS tables, consider only looking at the query text. Looker does stuff like create table database.schema.<looker_unique_identifier>_<table_name> so we don't currently group those together.
• Strip out literals used in where predicates ( select * from table where user_id = $1). These parameterized queries are common in dashboards and should be grouped together.

As discussed previously with @NiallRees , we want to give users the option to choose between which method they use to group together queries, so these should be kept as separate fields to ensure this flexibility remains.