Minor PlaidML fixes

pydocstyle 4.0 breaks flake8-docstrings.

This came up while debugging memory corruption in the ngraph-tf tests: plaidml_ops_replicate was holding onto
a reference to a temporary that was destroyed while the reference was still live.  Replacing that code with a
call to get_argument() eliminated the temporary and fixed the issue.  And in general, there's no reason to be
allocating temporary vectors when we're pulling out a single argument, so this change replaces the other
sequences of get_arguments().at(), too.

The PrefixReshapeElimination pass doesn't work correctly if the network's operation's input and output shape
compatibility is being validated; an implicit broadcast operation is required to get the shapes right.  This
change moves the implementation back to PlaidML (since it's seeming less generally useful, and PlaidML has an
implicit broadcast operation), and fixes both it and the implicit broadcast pass to only apply when the
downstream operation is an elementwise operation (since PlaidML doesn't support automatic broadcasting for
contractions).

Update setup.py

Improve PlaidML backend option debugging

Decomposition of fused ops in PlaidML backend [r0.23]

Add PlaidML Quantize & Dequantize

Added a missing guard that is breaking the build

Adjust PlaidML test tolerances

* Add LpNormalization operator along with unit tests.

* Add validation macro based on NGRAPH_CHECK.

* Add InstanceNormalization operation along with unit tests.

* Update supported ops table.

* Fix merge error.

* add send recv file

* update CMakeLists

* add to op_tbl, add to serializer

* fix bug in serializer, add MPI send recv

* not supported in MlSL

* GPU not support send recv

* add reference implementation

* send recv not supported by intel gpu

* resolve Scott's comment about construction

* style

* add comments, add to gpu not supported list

* using Output<Node> instead of shared_ptr<Node>

* add test

* disable test for MLSL

* disable test for cpu, gpu, gpuh

* add static string s_manifest = ;

* using NGRAPH_TEST so test can be disable for specific backend

* float number

* revert last change

* default

* support more data type

* change License text

* change License text

* add and ring send recv test

* add and ring send recv test

* skip send_recv_ring

* fix bug

* Run fused op decomposition recursively until no more fused ops.

* Update callback member name.

* Remove redundant conditions.

* Style apply.

fix license header again

MLIR Integration into CPU backend