Rethink interval puzzle generation

[Buduf]

So in the EarTrainer class we generate puzzles with different intervals and notes depending on the game difficulty. First we have to generate a random Note

On the easiest difficulty the first note can be only C, and then the second note which is either the same or one seminote higher. That's how it's implemented now and other difficulties similarly. So I think it's easier to guess from small intervals and big intervals. For example, I think it's easier to hear the difference between P4 and P8, than m2 and M2.

I suggest that we store the possible intervals in some collections for different difficulties. Maybe return the colleciton from a method. And then get a random element from this collection for the puzzle.

Also, we have to randomize the first note on higher difficulties.

I'm talking about this part of the code: https://github.com/Buduf/PTI-Ear-Trainer/blob/02153b460fcc0eedf989ea65adbc8920d8bb892b/PTI-Ear-Trainer-Model/EarTrainer.cs#L31-L48

[Buduf]

I hope I'm understandable.

[Buduf]

Here's a table of possible values for different difficulties.	Difficulty	Possible Notes	Possible Intervals
VERYEASY	C	P1, M2, P5, P8
EASY	LowerA, C, E	P1, M2, M3, P4, P5, P8
MEDIUM	LowerA, C, E	P1, m2, M2, m3, M3, P4, P5, M6, M7, P8
HARD	LowerA, LowerB, C, D, E, F	[all of them]
VERYHARD	LowerA, LowerASharp .. F	[all of them]

[Andre2220]

Okay, It's nice, I will start to implement this.