Disable exec checks for device_uvector::operator=

[miscco]

We already suppress those for the constructors

[wence-]

Makes sense to me, thanks

[bdice]

We had the same change proposed in #1786 but we don’t have any tests for this case. Can you share how you found this, and help us minimize it into a test so we don’t miss this anywhere else?

[miscco]

That was a fallout from building rapids, where rmm::device_vector was stored in a rmm::device_uvector

[bdice]

Docs builds are failing:

 /__w/rmm/rmm/python/rmm/docs/librmm_docs/data_containers.rst:4: WARNING: Error when parsing function declaration.
If the function has no return type:
  Error in declarator or parameters-and-qualifiers
  Invalid C++ declaration: Expecting "(" in parameters-and-qualifiers. [error at 23]
    RMM_EXEC_CHECK_DISABLE device_uvector & operator= (device_uvector &&) noexcept=default
    -----------------------^
If the function has a return type:
  Error in declarator or parameters-and-qualifiers
  If pointer to member declarator:
    Invalid C++ declaration: Expected '::' in pointer to member (function). [error at 38]
      RMM_EXEC_CHECK_DISABLE device_uvector & operator= (device_uvector &&) noexcept=default
      --------------------------------------^
  If declarator-id:
    Invalid C++ declaration: Expecting "(" in parameters-and-qualifiers. [error at 38]
      RMM_EXEC_CHECK_DISABLE device_uvector & operator= (device_uvector &&) noexcept=default
      --------------------------------------^

I'd also like to get a test that reproduces the previous failure into this PR if possible.