perf: remove unnecessary cloning and iteration

[jeswr]

This PR:

1. Fixes a bug where cached document contexts were being mutated if they were being loaded via in @import statement (fix: deep copy imported contexts #71 was originally opened to resolve this).
2. Removes all deep cloning to improve performance (perf: no deep cloning #72 was originally opened to resolve this)
3. Skips applying idempotent operations to the parent.

All unit & spec tests in the jsonld-streaming-parser pass with these changes (tested against bce9e05).

I have also done a (small) benchmark of these changes. As can be seen the time taken to parse a context with these changes significantly decreases (in particular it is now 20x faster to parse a VC context).

Before:
Parse a context that has not been cached; and without caching in place x 78.17 ops/sec ±0.73% (78 runs sampled)
Parse a context object that has not been cached x 2,280 ops/sec ±0.65% (91 runs sampled)

in #72:
Parse a context that has not been cached; and without caching in place x 123 ops/sec ±0.78% (84 runs sampled)
Parse a context object that has not been cached x 5,417 ops/sec ±0.70% (87 runs sampled)

After (in this PR):
Parse a context that has not been cached; and without caching in place x 1,714 ops/sec ±0.78% (84 runs sampled)
Parse a context object that has not been cached x 8,902 ops/sec ±0.70% (87 runs sampled)

Let's get this merged and then I'll revisit #70 to see what is actually worth caching.

[coveralls]

Pull Request Test Coverage Report for Build 6706243153

• 53 of 53 (100.0%) changed or added relevant lines in 2 files are covered.
• No unchanged relevant lines lost coverage.
• Overall coverage remained the same at 100.0%

---

Totals
Change from base Build 6492290900:	0.0%
Covered Lines:	553
Relevant Lines:	553

---

💛 - Coveralls

[rubensworks]

Nice optimization work @jeswr!

[rubensworks]

Impressive work @jeswr!
Released as 2.4.0.