[Illviljan]

Improve performance by removing type and shape checking in _protect_dataset_variables_inplace by setting directly to self._data.
The function already has valid variables, there should be no need to recheck when adding small wrapper classes like CopyOnWriteArray and MemoryCachedArray.

variable.data = data does this, and as_compatible_data is a known performance bottleneck:

xarray/xarray/core/variable.py

Lines 452 to 456 in 447e5a3

	@data.setter
	def data(self, data: T_DuckArray | ArrayLike) -> None:
	data = as_compatible_data(data)
	self._check_shape(data)
	self._data = data

Add some typing to make sure strange types are not input to these functions.

• Related to opening datasets slow compared to netCDF4 #9058
• Tests added
• User visible changes (including notable bug fixes) are documented in whats-new.rst
• New functions/methods are listed in api.rst

[Illviljan]

Found 2 errors in 1 file (checked 171 source files)
xarray/backends/api.py:241: error: Incompatible types in assignment (expression has type "MemoryCachedArray", variable has type "_arrayfunction[Any, Any] | _arrayapi[Any, Any]")  [assignment]
xarray/backends/api.py:243: error: Incompatible types in assignment (expression has type "CopyOnWriteArray", variable has type "_arrayfunction[Any, Any] | _arrayapi[Any, Any]")  [assignment]

And here's the rabbit hole... :(
I suspect it probably is not as easy to just add some methods to these classes.