fix loading `in-memory` catalog (apache#1725)

Previously `CatalogType` was fetched by key,
`CatalogType[provided_catalog_type.upper()]` (note the bracket)
This PR changes `CatalogType` to be fetched by value,
`CatalogType(provided_catalog_type.lower())` (note the parenthesis)


https://stackoverflow.com/questions/29799235/python-enum-value2member-map-accessor

This fix loading `in-memory` catalog, `load_catalog("catalog",
**{"type": "in-memory"})`.
Previously, `"in-memory"` caused a key error because its key is
`IN_MEMORY`. Note the `-` vs `_`.


https://github.com/apache/iceberg-python/blob/1d24e71041e35e26e126aa4508ed7384e8aa031c/pyiceberg/catalog/__init__.py#L113-L119


We want `in-memory` and not `in_memory` to match spark

https://github.com/apache/iceberg-python/blob/1d24e71041e35e26e126aa4508ed7384e8aa031c/dev/spark-defaults.conf#L35

PyIceberg 0.9.0rc3

PyIceberg 0.9.0rc2

PyIceberg 0.9.0rc1

bump to 0.8.1

PyIceberg 0.8.1rc1

Bug Fix: `metadata_location` to be optional in `TableResponse` (apach…

…e#1321)

* make metadata_location optional

* add test for staged creation

* revert

* revert from testing

PyIceberg 0.8.0rc2

PyIceberg 0.8.0rc1

PyIceberg 0.7.1