Add arbitrary shard-grid core range set support to addrgen in command processor (ttnn/cpp/ttnn/operations/ccl/common/kernels/ccl_send_reader_two_input.cpp)

[SeanNijjar]

For TG Llama integration, it's required that CCLs support arbitrary shard grid configurations.

Support for this can be added by updating both generate_multi_command_stream_kernel_rt_args to pass runtime args for the sharded addrgen and generate_multi_command_stream_kernel_ct_args for compile time args, respectively.

For all CCL's to enable this behaviour from host side, I think only these should require updates.

Required Test Cases:

• input and output shapes
• input and output shard specs
• other op specific attributes (e.g. gather dim)

TODO

[SeanNijjar]

FYI @johanna-rock-tt for arbitrary shard grid support. Please add test cases here (including, if possible, the padded shard cases that were newly brought up today).