UUID no longer inherits from string.  It is still possible to cast
between string and UUID but they are not strict descendant/ancestor.

BREAKING CHANGE: Any code that was relying implicitly on string-y
behavior from UUID datatypes will need to add an explicit cast first.

add support array index op

add method for converting of dtype to typename

add support array concat op

update array index op

it uses translate of exprs

fix format

add support array repeat op

add support array slice op

add DateTime64 to ch dtypes

update test_array_index for check negative ids

add support for neg slice ids

use f-string in array repeat op

refactor to_ibis method

style: oneline expr for array repeat

Co-authored-by: Phillip Cloud <417981+cpcloud@users.noreply.github.com>

test: remove xfail on array slice and index ops

removed for:
* dask
* pandas
* postgres
* pyspark

fix: add negative index support for postgres

style: remove unused module

These scripts are no longer used.

- remove explict repr values from tests

These intentionally collide with the jobs in `ibis-main.yml` and
`ibis-backends.yml` and run on the inverse of the path skipping logic
present in those two files.

This should allow CI to finish and all jobs to return, even if they're
skipped.

This is an ugly hack, but is the supported one for now.

Because they are of debatable utility.

…ames in relations

…spark backend

The Pandas and Dask backends always pass scope to adjust_context, so
that adjust_context can potentially make use of extra information in
scope. The PySpark backend was not passing scope to adjust_context,
which can lead to behavior differences or errors across backends.

fixes #3108