Resolve II and Latency issues

Created by: thesps

Reviewing #280 I saw that the latency and II of some of the test models is not as expected. e.g. the model [1] with RF=2 was synthesizing with II=19, and the model [2] with RF=112 synthesized with II= 117-min, 118-max. You can see this in the reports of #280. Basically, with Strategy: Resource, some functions that should have been pipelined were not. These additional pragmas resolve that. Regarding the inline pragmas in the dense- and softmax-wrapper functions: those were automatically inlined in the Strategy: Latency tests, and only the wrapper function is inlined, not the actual dense or softmax implementation underneath.

I've marked the PR WIP since I still see a different min/max latency for the Strategy: Resource which we should also resolve. Although we could take these changes sooner if that one takes too long to find.

[1] KERAS_3layer-xcvu9p_flgb2104_2_e-c5-io_parallel-rf2-ap_fixed_16_6_-Resource-2020.1 [2] qkeras_mnist_dense-xcvu9p_flgb2104_2_e-c5-io_parallel-rf112-ap_fixed_16_6_-Resource-2020.1