indicator chaining

[DaveSkender]

Description

Resolve #630 and adding general indicator chaining so you can do things like:

# choose any candle part or variant
# to override the default
var results = quotes
  .Use(CandlePart.OHLC4)
  .GetRsi(14);

# slope of any indicator
var results = quotes
  .GetMacd()
  .GetSlope();

Note: not all combinations are available as a secondary combination since some require the full quote for calculation (e.g. ATR, Heikin-Ashi require multiple candle parts).

Breaking changes

• numerous result values and some parameters changed from decimal to double
• result class is now sealed and cannot be used as a derived class (required for reusable chaining), though you can just replicate the class properties
• IMPORTANT!: Beta and Price Relative Strength (PRS) have swapped positions of Eval and Base (Market) quote parameters positions, for consistency. Since these have the same data type profile, it does not naturally produce a warning! A warning is forced on v2.0.0 only as a precaution but will be removed in the next released version.
• GetSmaExtended() was renamed to GetSmaAnalysis(). We'll expand on and improve extended analysis broadly in future versions. See Extended Analysis #320 on how this may change in the future.
• results now return NaN instead of null in random div/0 scenarios; though, we intend to revert this change to reduce user friction, see random NaN values #839 --> please note, this was reversed in v2.0.3 and now returns null

[LeeDongGeon1996]

This is going to be a giant step. I need to think about how to apply these changes to Python wrapper 😂(especially, indicator chaining)

[DaveSkender]

This is going to be a giant step. I need to think about how to apply these changes to Python wrapper 😂(especially, indicator chaining)

@LeeDongGeon1996, I know. It was a more significant thing than expected. It's likely going to be a major v2.0 release change.

The old APIs all still work just fine, so there's nothing major that should break its use in the Python wrapper, with the exception of a few noted things above, and there's certainly some new features to add. Most of the work in this PR is just refactoring code to make it more maintainable and to prepare for some future stuff (e.g. streaming uses cases).

It might be better to get the wrapper to a stable 1.0 before starting on some of this.

[LeeDongGeon1996]

I've thought about the way to add these features(chaining, streaming) and figured out that it needs substantial refactoring.

It might be better to get the wrapper to a stable 1.0 before starting on some of this.

I think so. And if possible, I'd like to version up pythonnet so that the library can support .NET 6.0. :)