Reduce CPU usage for Deserializer

[agnosticdev]

Change to help optimize generic operations on the Deserializer both internally and across module boundaries.
Running our Deserializer benchmark that creates 1 frame and runs the Deserializer on it 10,000,000 times I see this change cut the CPU usage in half.

Currently, top of tree:

434.41 M 89.1%	52.65 M	      specialized static Deserializer<>.deserialize<>(_:claim:_:)	
298.24 M 61.1%	13.00 M	       static Deserializer<>.deserialize<>(_:_:)	
262.99 M 53.9%	12.00 M	        specialized static Deserializer<>.deserialize<>(_:_:)	
191.63 M 39.3%	13.50 M	         partial apply for closure #1 in runtest()	
178.12 M 36.5%	25.97 M	          closure #1 in runtest()	
74.63 M  15.3%	27.44 M	           Deserializer<>.uint8(_:)	
27.00 M  5.5%	-      	           Deserializer<>.uint64NetworkByteOrder(_:)	
23.67 M  4.9%	-      	           Deserializer<>.uint16NetworkByteOrder(_:)	
14.54 M  3.0%	-      	           Deserializer<>.uint32NetworkByteOrder(_:)	
12.32 M  2.5%	12.32 M	           type metadata accessor for Deserializer<EmptySpanFactory>	
19.36 M  4.0%	19.36 M	         Deserializer<>.uint64NetworkByteOrder(_:)	
15.00 M  3.1%	15.00 M	         specialized Deserializer<>.finalResult.getter	
9.00 M   1.8%	9.00 M 	         Deserializer<>.init<>(_:)	
9.00 M   1.8%	9.00 M 	         Deserializer<>.uint32NetworkByteOrder(_:)	
7.00 M   1.4%	7.00 M 	         Deserializer<>.uint16NetworkByteOrder(_:)	
22.24 M  4.6%	22.24 M	        partial apply for closure #1 in runtest()	
83.53 M  17.1%	72.53 M	       Frame.bytes.getter	
11.00 M  2.3%	11.00 M	        Frame.startOffset.getter	

And with these changes:

217.45 M 87.5%	3.74 M 	      static Deserializer<>.deserialize<>(_:claim:_:)	
136.51 M 54.9%	21.00 M	       specialized static Deserializer<>.deserialize<>(_:_:)	
98.91 M  39.8%	3.42 M 	        partial apply for closure #1 in runtest()	
95.49 M  38.4%	-      	         closure #1 in runtest()	
42.62 M  17.1%	42.62 M	          specialized Deserializer<>.uint8(_:)	
19.24 M  7.7%	19.24 M	          specialized Deserializer<>.uint32NetworkByteOrder(_:)	
17.81 M  7.2%	17.81 M	          specialized Deserializer<>.uint16NetworkByteOrder(_:)	
15.81 M  6.4%	15.81 M	          specialized Deserializer<>.uint64NetworkByteOrder(_:)	
9.60 M   3.9%	9.60 M 	        Deserializer<>.init<>(_:)	
7.00 M   2.8%	7.00 M 	        specialized Deserializer<>.finalResult.getter	
57.63 M  23.2%	49.63 M	       Frame.bytes.getter	
19.57 M  7.9%	19.57 M	       partial apply for closure #1 in runtest()	
4.00 M   1.6%	-      	      specialized IndexingIterator.next()	

So as you can see these changes really help the generic functions optimize better.