Improve dbt seed times by removing type lookup and casting during INSERT

[nrichards17]

Resolves #476

Description

Several users have noticed slow run times for loading dbt seeds with >1k records when using the dbt-databricks adapter, with run times becoming prohibitively slow for seeds with >10k records. This PR speeds up that run time substantially by removing unnecessary type lookup and casting during the INSERT statement.

DBT seeds essentially are built in two steps:

1. The table is created with the appropriate column types (explicit or inferred) with CREATE TABLE AS ...
2. The values are loaded into that table with INSERT OVERWRITE INTO table.schema VALUES ...

For some reason in the second step, there is both a type lookup and subsequent CAST(x) AS type for every single value (rows x columns) in the seed. This is effectively redundant and unnecessary, since the type information was already used when defining the column types during table creation.

Removing these steps significantly speeds up the seed run times. For example, I was able to load a seed with 47k records and 9 columns in about 1 minute with this change, whereas previously the seed hadn't even finished after 10+ minutes of loading.

Checklist

• I have run this code in development and it appears to resolve the stated issue
• This PR includes tests, or tests are not required/relevant for this PR
• I have updated the CHANGELOG.md and added information about my change to the "dbt-databricks next" section.

[benc-db]

@nrichards17 I'm on vacation next week, but I'll run this change against our test suite when I get back, and assuming it passes, will try to sneak it in before I release 1.7.0.

[benc-db]

Thanks for the PR, appreciate it.

[nrichards17]

happy to help, thank you @benc-db !

[benc-db]

Closing in favor of 498 which can run against our infra. Thanks @nrichards17!