feat: Implement DFSchema.print_schema_tree() method

[comphead]

Which issue does this PR close?

• Closes Create schema print out method #17460.

Rationale for this change

What changes are included in this PR?

This PR adds a new print_schema_tree() method to DFSchema that formats schema information in a tree-like structure similar to Apache Spark's schema display format, with proper nested indentation for complex data types.

Core Implementation
File: datafusion/common/src/dfschema.rs

• Added print_schema_tree() method
  Public method that formats the entire schema in tree structure
  Handles both qualified and unqualified field names
  Returns formatted string with "root" header and proper indentation

Added format_field_with_indent() helper function

Recursive function that handles nested indentation for complex types
Supports proper tree structure with |-- and | indentation
Handles all Arrow data types

Example output for map of array

root
 |-- array_map_field: list (nullable = false)
 |    |-- item: map (nullable = false)
 |    |    |-- key: string (nullable = false)
 |    |    |-- value: string (nullable = false)

Are these changes tested?

Are there any user-facing changes?

[comphead]

this is unrelated change, just removing unnecessary loop

[alamb]

It seems to me like this change just moves the call to normalize the identifier into the map function (I don't think there are any loops removed 🤔 )

[comphead]

it was 2 map before, I hope rustc is smart enough to merge them in runtime, but in this case it slighlty more readable as well.

[alamb]

I think the change is fine, I just wanted to make sure I understood what was going on. Thank you @comphead

[alamb]

Thank you @comphead -- I have some suggestions, but nothing that I think would prevent this PR from merging

BTW I was wondering if this would be a better default Display for DFSchema, but it seems like there is already a default implementation

[alamb]

using insta might make these cases easier to update / easier to see. Something like this perhaps

datafusion/datafusion/core/tests/user_defined/user_defined_aggregates.rs

Lines 214 to 220 in b8bf7c5

	insta::assert_snapshot!(batches_to_string(&actual), @r###"
	+---------------------------------------+
	| sum(arrow_cast(t.time,Utf8("Int64"))) |
	+---------------------------------------+
	| 19000 |
	+---------------------------------------+
	"###);

[alamb]

i think an insta snapshot would be better here

[alamb]

It seems to me like this change just moves the call to normalize the identifier into the map function (I don't think there are any loops removed 🤔 )

[comphead]

BTW I was wondering if this would be a better default Display for DFSchema, but it seems like there is already a default implementation

Display provides more information now, like metadata, dict_ordering. Not sure if it is a right time to replace but in future why not.

To display tree schema we need to ship 1 more DDL function to show the schema in the CLI or by calling DataFusion sql.
Alternatives what DDL can be chosen are #17466, I'm planning to make a vote on this

[alamb]

I think the change is fine, I just wanted to make sure I understood what was going on. Thank you @comphead