OpenCL C lacking a unique identifier

Consider adding to the specification, a function that will allow each work group (that is currently being executed on a compute unit) to get some kind of compute id or unique id("get_compute_id()" or "get_unique_id()") that no other work group can use until it has finished its execution and releases the ID to be reused.

e.g.
//be able to get the highest possible index in the host code
device.getInfo<CL_DEVICE_MAX_NUM_WORKGROUPS_THAT_CAN_BE_EXECUTED_AT_A_TIME>() == 19 //assuming the device has 20 compute units and ONLY ONE work group can be executed per compute unit at a time; otherwise it would be higher than num compute units

//in the kernel, be able to get the unique id for the current work group
int uniqueID = get_unique_id(); //return int ranging from 0 to CL_DEVICE_MAX_NUM_WORKGROUPS_THAT_CAN_BE_EXECUTED_AT_A_TIME

Start by reviewing the OpenCL C specification and the host device-info example described in the issue. Clarify the scope and semantics of a work-group identifier, including its lifetime, range, and interaction with concurrent work groups. Done would require an agreed specification change rather than only an implementation edit.