Remove SoftmaxQuantizer

Summary:
Softmax not supported on current platforms.

This introduces quant/dequants into the graph, which are unneeded.

This diff removes the quantizer until supported.

Differential Revision: D81964057

Author: JakeStevens

backends/nxp/quantizer/neutron_quantizer.py

@@ -225,7 +225,6 @@ def __init__(self):
                 NeutronAtenQuantizer(ReluInPlacePattern(), static_qconfig),
                 NeutronAtenQuantizer(ReshapePattern(), static_qconfig),
                 NeutronAtenQuantizer(SigmoidPattern(), static_qconfig),
-                NeutronAtenQuantizer(SoftMaxPattern(), static_qconfig),
                 NeutronAtenQuantizer(TanhPattern(), static_qconfig),
                 NeutronAtenQuantizer(TanhInPlacePattern(), static_qconfig),
                 NeutronAtenQuantizer(ViewPattern(), static_qconfig),