feat: resolving invalid parsing causing stack overflow

[galkahana]

This is intended as resolution to #278 and #279.
Both issues report test files that when parsed cause a stack overflow.
The stack overflow is set up by providing many multiple array start char - [ - or dict start chars - << . Each such start triggers a new recursion level at the object parser looking to parse the items of the array or dict. when in themselves they are arrays or dicts (by having that starter token) there's another level...and another...and another.

I put a max depth, of either dicts or arrays parsing at 100. This includes resolving cases of such sequences that intermingle array and dict start like <<[<<[<<[[[[<<[<<<<[. If there's a depth of 100 of either dicts or arrays including each other the code will halt parsing with a failure and return. This limit should take any valid PDF and parse it well, and stop naughty PDFs.