jsonchema: add memoization and cycle detection (#77)

[albertsundjaja]

This PR adds:

1. Cycle detection: return error instead of waiting for stack overflow*
2. Memoization: performance improvement**

*This is a fix for #77
The root cause of the issue was the type being processed is deeply nested and has cycle. Hence, the heap memory growth is faster than the stack growth, it went OOM before stack overflow panic can kick in

**I posted a perf test in the issue thread #77
Copying it here for visibility

// Without memoization
1. Testing SimpleStruct:
Running 10000 iterations...
Performance: 22.360083ms for 10000 iterations
Average per iteration: 2.236µs

2. Testing NestedStruct:
Running 10000 iterations...
Performance: 79.656125ms for 10000 iterations
Average per iteration: 7.965µs

// Memoization with deep copy -- implemented in this PR
1. Testing SimpleStruct:
Running 10000 iterations...
Performance: 14.800583ms for 10000 iterations
Average per iteration: 1.48µs

2. Testing NestedStruct:
Running 10000 iterations...
Performance: 50.574417ms for 10000 iterations
Average per iteration: 5.057µs

// With memoization using gob
1. Testing SimpleStruct:
Running 10000 iterations...
Performance: 515.872416ms for 10000 iterations
Average per iteration: 51.587µs

2. Testing NestedStruct:
Running 10000 iterations...
Performance: 646.161875ms for 10000 iterations
Average per iteration: 64.616µs

arguably, the benefit of the memoization is minimal. In addition, in the case of the caller invokes For with a struct that contains a cycle, some of the fields in that struct might be cached which incur unnecessary memory cost. I'm happy to remove the caching and just keep the cycle detector

[jba]

Thanks. I agree, let's drop the caching and keep the cycle detection.

[jba]

Let's also change this to "It also returns an error if there is a cycle in the types."