This reverts merge request !9

not). Code generation uses it.
- Added a policy that uses it.

Tested on pipeline benchmark.

(benchmark not updated)

Bug fix in nvptx and spir backends related to adding arguments to
node functions.

(CPU/GPU/SPIR) based on policy.
Support for scheduling per iteration included.
Further testing required.

Made visc-rt threadsafe by implementing locks across all ocl functions. Very crude. Fine grained locking possible if performance deteriotes

    - IntrinsicsVISC.td, visc.h, GenVISC.cpp
(2) Simplified GenVISC to easily add support for new intrinsics which have a 1
to 1 mapping instruction during code gen
(3) Added runtime api call to set ocl shared memory argument

request_mem failure
(2) modified visc_sh version of sgemm