Pseudo-instructions need better representation

[AFOliveira]

Is your feature request related to a problem? Please describe.
Pseudo-instructions are represented in a way that is not ideal to represent it's whole extent. For instance they only capture two informations:

when: 
to:

Describe the solution you'd like
I think we should have a per-yaml representation of each pseudo-instructions that refers to it's parent instruction - rather than the opposite.

Additional context
There has been discussion and a previous try at this on #301

[ThinkOpenly]

Note that the mapping from pseudoinstruction to instruction is not always 1-to-1.
From RISC-V Assembly Programmer's Manual:

[AFOliveira]

Yeah, even gets worse than that since if I remember correctly li can turn into 5 different instructions and there are conditional pseudo-instructions that represent different instructions based on what extensions are implemented right?

As for the expansion I think we can have something like a to: field in which we represent two instructions - just need to the backends to handle it. However, I am not sure on how to represent the condition.

[dhower-qc]

dusting off this topic that keeps resurfacing, I think we determined:

• an alias is a pseudo-instruction with a 1:1 mapping that does the same thing as the original encoding. We encoding those in UDB using the pseduoinstructions key in an instruction object.
• an override is an instruction that uses the encoding space of another instruction, but does something different. Currently, this is marked in UDB with the hint key on the parent instruction (which should be renamed; originally we were just using it for hints) and the override as a stand-alone instruction object.
• a series is a pseduo-instruction with a standard N:1 mapping. We aren't currently encoding these.
• we aren't going to try to represent li. We might decide to add it to the DB, but without any attempt to representing a mapping.

[AFOliveira]

@dhower-qc we have definitely agreed on all of those and I should have mentioned those conclusions for sure, but pardon my ignorance when I ask if those things are already implemented in UDB? I don't think alias and override are, right?

[dhower-qc]

alias and override are implemented, just with different names. alias == pseudoinstructions on the instruction definition. override == hint on the instruction definition. series is not implemented.