fix(tsql, postgres)!: Improve UUID support

[amachanic]

This PR addresses some UUID support issues with T-SQL and Postgres. A small example of what I wanted to get working:

Input: sqlglot.transpile('create table x (i uniqueidentifier not null default(newid()))', read='tsql', write='postgres')
Output: ['CREATE TABLE x (i UNIQUEIDENTIFIER NOT NULL DEFAULT (NEWID()))']
Expected: ['CREATE TABLE x (i uuid NOT NULL DEFAULT (gen_random_uuid()))']

• Removed unnecessary UNIQUEIDENTIFER token (standardized on UUID)
• Added NEWID() support to T-SQL
• Cleaned up Spark UUID support, which I noticed was also broken. (I think Hive might be, too, but I don't know for a fact so I didn't touch that one.)

[georgesittas]

Thanks for the contribution! FYI, this is breaking, so the PR title needs to reflect that. I'll make the change, just something to keep in mind for future PRs.

Looks good 👍