Add create index and drop index support

[miuy56dc]

This PR adds support for:

CREATE [ UNIQUE ] INDEX [ IF NOT EXISTS ]
    <index_name>
    ON <table_name>
    ( col_name [, ...] )

DROP INDEX <index_name>

[nickolay]

Thanks for the PR, and sorry it took me so long to get back to you!

I see from the tests this supports CREATE UNIQUE INDEX IF NOT EXISTS idx_name ON test(name) and DROP INDEX idx_name. This agrees with Postgres' syntax and has an intersection with MSSQL's syntax, which is good.

However the code in parse_index_column() is more complex than what's required to support the above, and I don't understand what specifically it tries to implement. A simple self.parse_comma_separated(Parser::parse_identifier)? should be able to parse the above example. self.parse_comma_separated(Parser::parse_order_by_expr) seems like it would support all but the most exotic syntax.

Did you intend to support trailing comma? (let's make it explicit if so, and I would prefer introducing a helper akin to parse_comma_separated instead of duplicating code from parse_columns)

Why do you parse an individual item of an index definition as a sequence of identifiers, delimited by whitespace only, and store it as an ObjectName (which represents a dot-delimited name like foo.bar.baz)? This seems wrong and wouldn't round-trip when serialized back to string...

A few minor quibbles below, which I would've fixed myself if it weren't for the above questions.

[miuy56dc]

@nickolay Thanks for the comment. And yes I should use self.parse_comma_separated(Parser::parse_identifier)? to parse column names.
Actually I don't intend to support trailing comma, just because I find parse_columns support it, so I do the same thing.

[nickolay]

Alright, supporting CREATE INDEX idx_name ON test (col_name[, ...]) sounds great to me!

In this case, there's no reason to have parse_index_column as a separate method (it's rather short now!), this would lead to changing let object_names into let columns, as expected by Statement::CreateIndex. If you can make this last change and squash the changes into a single commit rebased on top of the current master, I'll merge it now. Or I can take it from here, if you prefer.

[miuy56dc]

Hi @nickolay, thanks for the advice, I have moved parse_index_column into parse_create_index and squash the changes into a single commit rebased on top of the current master.

[nickolay]

Excellent, thanks so much for bearing with me and for taking it to the finish line!

I just realized this can now use parse_parenthesized_column_list, but as I wanted to credit you in the CHANGELOG anyway, I've gone ahead and made the last minute change myself.