fix(spark): align parse_url empty FILE path

[kumarUjjawal]

Which issue does this PR close?

• Closes [datafusion-spark] parse_url incompatibilities #21943.

Rationale for this change

parse_url in the Spark function library did not match Spark for some empty path URL cases. In particular, absolute URLs with no explicit path were treated as if they had / because the URL parser normalizes the path.

Also handles

• parse_url(..., 'PATH') now returns / for an explicit root path like http://example.com/.
• parse_url(..., 'QUERY', key) now returns the raw query value, so percent-encoded values like x%20y are not decoded to x y.

What changes are included in this PR?

This PR keeps the existing URL parser, but adjusts FILE extraction so parser-normalized / is treated as empty only when the original URL had no path after the authority.

It also adds regression coverage for parse_url and try_parse_url, including the boundary between a missing path and an explicit root path.

Are these changes tested?

Yes

Are there any user-facing changes?

parse_url and try_parse_url now match Spark more closely for empty path FILE results. There is no public API change.

[kumarUjjawal]

There is another issue which should be part of this pr, I'm working on the fix.