Benchmarks for context conversion

[anuptalwalkar]

Type conversion benchmarks for wrapping context.Context with fx.Context struct vs using Upgrade/Convert function with type assertion.

Looks like with type assertion when no WithValue is called, saves us 1 alloc and ~7 ns/op. When WithValue is called, type assertion is 8 ns/op expensive.

bash-3.2$ go test -bench=. -benchmem
BenchmarkContext_StructWrapper-8                	30000000	        55.3 ns/op	      32 B/op	       3 allocs/op
BenchmarkContext_StructWrapperWithValue-8       	30000000	        43.5 ns/op	      32 B/op	       3 allocs/op
BenchmarkContext_WithTypeAssertion-8            	30000000	        47.8 ns/op	      16 B/op	       2 allocs/op
BenchmarkContext_WithTypeAssertionWithValue-8   	20000000	        52.8 ns/op	      32 B/op	       3 allocs/op

[coveralls]

Coverage remained the same at 91.717% when pulling 425ef67 on context-benchmark into aef180e on master.

[madhuravi]

What is the conclusion for what we want to do based on this?

[madhuravi]

It's confusing that the test StructWrapper is calling the function typeConversion on line 41 whereas the test WithTypeAssertion is calling the function typeUpgrade. Fix and clarify naming.

[ascandella]

Numbers look good. Nanoseconds are totally fine. Allocs seem small. I don't think we need to worry about optimizing this for now.