Added num_keypoints arg to TfExampleDecoder call

[dustindorroh]

Added num_keypoints arg to TfExampleDecoder call in dataset_builder.py

This arg is already part of the proto spec:

models/research/object_detection/protos/input_reader.proto

Lines 89 to 90 in b9ef963

	// Number of groundtruth keypoints per object.
	optional uint32 num_keypoints = 16 [default = 0];

The default value in TfExampleDecoder is correctly mapped.

models/research/object_detection/data_decoders/tf_example_decoder.py

Lines 136 to 147 in b9ef963

	class TfExampleDecoder(data_decoder.DataDecoder):
	"""Tensorflow Example proto decoder."""
	def __init__(self,
	load_instance_masks=False,
	instance_mask_type=input_reader_pb2.NUMERICAL_MASKS,
	label_map_proto_file=None,
	use_display_name=False,
	dct_method='',
	num_keypoints=0,
	num_additional_channels=0,
	load_multiclass_scores=False):

I believe this may be relevant in #2366 and #4676