An attempt to improve performance of terminal/parser

# Description of the new feature/enhancement

Currently `StateMachine` use many if-else branches which may not an efficient way to implement a state machine. It could be improved by using a table approach.

# Proposed technical implementation details (optional)
A quick and dirty implement lays on https://github.com/flyingcat/terminal/tree/vtparser
* v1
* Use `codegen.ps1` to generate a function pointer table to execute state transition.
* Only need to modify `stateMachine.hpp|cpp`. It has a demo build on [vtparser_apply ](https://github.com/flyingcat/terminal/tree/vtparser_apply) branch.
* v2
* Merge `OutputStateMachineEngine` and use CRTP to reduce virtual calls, as the separation seems only for test and actually no need for runtime dispatch.
* Instead of using switch-case to dispatch actions on `ActionExecute` and `ActionCsiDispatch`, directly dispatch them on corresponding char.
* Unfortunately performance improvement is not as much as I expected and it affects many code. So may not worth it.

Some surprise:
* Indirect functions calls are slower on x64 than ARM64. A possible optimization is eagerly pulling chars if easy branch prediction like `_ActionParam`.
* On ARM64, the cost of tracing is heavy. Remove `TraceOnEvent` and `TraceCharInput` will reduce a lot of time.

Benchmarks (*_P for plain text, *_V for nvim vt output):
```
x64 ARM64 ARM64 no heavy tracing
------------------------------- ------------------------------- -------------------------------
> VT_EN_P : 2.56 MB Passed. > VT_EN_P : 2.56 MB Passed. > VT_EN_P : 2.56 MB Passed.
parser 0: 4907.81 us parser 0: 5061.98 us parser 0: 5052.18 us
parser 1: 4869.68 us, +0.8% parser 1: 5110.27 us, -0.9% parser 1: 4376.73 us, +15.4%
parser 2: 4450.93 us, +10.3% parser 2: 4330.88 us, +16.9% parser 2: 3803.58 us, +32.8%
> VT_EN_V : 11.36 MB Passed. > VT_EN_V : 11.36 MB Passed. > VT_EN_V : 11.36 MB Passed.
parser 0: 150354.27 us parser 0: 167648.47 us parser 0: 165199.61 us
parser 1: 104640.22 us, +43.7% parser 1: 122205.31 us, +37.2% parser 1: 88889.78 us, +85.8%
parser 2: 98848.33 us, +52.1% parser 2: 106223.27 us, +57.8% parser 2: 83710.12 us, +97.3%
> VT_CN_P : 2.12 MB Passed. > VT_CN_P : 2.12 MB Passed. > VT_CN_P : 2.12 MB Passed.
parser 0: 2746.15 us parser 0: 3009.11 us parser 0: 2982.51 us
parser 1: 2735.91 us, +0.4% parser 1: 3085.54 us, -2.5% parser 1: 2829.58 us, +5.4%
parser 2: 2613.93 us, +5.1% parser 2: 2759.70 us, +9.0% parser 2: 2793.15 us, +6.8%
> VT_CN_V : 11.29 MB Passed. > VT_CN_V : 11.29 MB Passed. > VT_CN_V : 11.29 MB Passed.
parser 0: 214276.04 us parser 0: 238064.93 us parser 0: 236772.92 us
parser 1: 146150.74 us, +46.6% parser 1: 175675.84 us, +35.5% parser 1: 127224.79 us, +86.1%
parser 2: 140112.63 us, +52.9% parser 2: 154611.41 us, +54.0% parser 2: 121005.83 us, +95.7%
```
```
Execute .\bc.exe -v -vc .\bc_data.txt

x64 ARM64
------------------------------- -------------------------------
before before
129.715MB, 9.118s, 14.227MB/s 129.713MB, 9.246s, 14.029MB/s
after after
129.713MB, 8.384s, 15.471MB/s 129.711MB, 8.327s, 15.578MB/s
```

----------

Sorry for the bad English and messy code. Hope the idea is clear enough.