Smarter replacement checks for columns

[MarkBerube]

Feature Request

• Yes, I reviewed the contribution guidelines.

Describe your use case and the problem you are facing
I have been actively trying to increase the velocity on my own URL S&Rs. To benchmark/diagnose how I could make them faster I started watching the SQL connections on large, test DBs with about 20gb worth of database rows living in wp_posts.

What I tended to notice are queries like the following:

UPDATE `wp_posts` SET `post_status` = REPLACE(`post_status`, 'http://dontmatter.co/', 'http://dontmatter.test/')

basically S&R is trying to replace on a column we can say confidently is not going to hold something I'd like to replace. One way around this is skipping these checks is to skip on this particular column. In my benchmarks case I skipped the following:

docker-compose run --rm wpcli wp search-replace --url=http://dontmatter.co/ 'http://dontmatter.co/' 'http://dontmatter.test/' wp_posts --recurse-objects --skip-columns=post_title,post_excerpt,post_status,comment_status,post_password,post_name,to_ping,pinged,guid,post_type,post_mime_type

a command that used to take me 32 minutes to execute now takes me 7 minutes with these changes.

Describe the solution you'd like

While I could do this for every single table in WP on my own, I feel this could be easier by being smarter about what columns we could skip automatically. Like we're never going to see something we'd like to replace on columns that act as binaries or columns that are like enums such as post_type that are all typed as varchar in the db.

[danielbachhuber]

Thanks for the suggestion, @MarkBerube !

I'm open to a PR with improvements, as long as it's backwards compatible.

Alternatively, introducing a dedicated search-replace command for URLs might be a more productive path forward: #186

I don't have a strong opinion of one vs. the other at this time.