feat: log parser speed improvements

[lcottercertinia]

Is your feature request related to a problem?

We want the parser to be as fast as possible especially to give head room for adding new features.

Describe the solution you'd like

1. Generator to reduce looping and heap usage
2. Only substring / slice when needed (remove String.split) (Creating new objects , string, array is expensive)
3. indexOf is our friend to avoid unnecessary string creation

Parser performance improvement from v1.12.1 for v1.14.0 pre release

Testing shows between 1.5 and 2.7 times faster
This is mainly down to

• using a generator to create the tokens on demand and keep memory low as well and looping lines only once. This reduces GC overhead which is expensive.
• cache namespaces to avoid having to re parse namespace strings which is expensive
• aggregate up the tree using a map for depth rather than recursion. Recursion is very very expensive.

[lcottercertinia]

2 to 3 times faster parser performance between v1.12.0 and v1.14.0 pre release.
That is so far. We expect a few more minor improvements before the release.