Fix possible overflow in WriteCoalescingCallbackWrapper::TearDown

[davidwendt]

Description

Fixes possible overflow in WriteCoalescingCallbackWrapper::TearDown function if the tile_out_count is sufficiently large enough. The out_char += blockDim.x could overflow when within block-size of the max of tile_out_count.

Checklist

• I am familiar with the Contributing Guidelines.
• New or existing tests cover these changes.
• The documentation is up to date with these changes.

[davidwendt]

This may not be likely since tile_out_count is unsigned so has a much larger max size and may not practically ever be significantly large. It was difficult to trace down what the possible value range is expected here.

[davidwendt]

/merge