Refactor quantization utils

See merge request !684

Fix bug: transform fc of tensorflow.

See merge request !683

deconv support variable output shape  and support Maximum op

See merge request !681

update gemmlowp

See merge request !682

add gemmlowp benchmark

See merge request !680

Fix tensorflow dsp converter

See merge request !679

* Add MacOS support in travis

* make changes according to the review

* change command order

* skip ops_test on mac

Fix openmp get core count bug

See merge request !677

add version script docs

See merge request !676

update generic library outputs and android demo

See merge request !674

package the output files after run.

See merge request !672

Fix bug: host dynamic library not work.

See merge request !673

fix convert caffe normal convolution failed when group == 1.

strip generated static lib

See merge request !671

fix lws overflow

See merge request !670

mv ISSUE_TEMPLATE.md as default

See merge request !668

Fix the bug: tuning 2d lws for gpu do not work.

See merge request !669

Support tensorflow fc layer.

See merge request !667

deconv support caffe and  crop op for caffe

See merge request !657

Calculate shape if it cannot be inferred

See merge request !666