fix: decode tree crash with nil safety and add missing Zicfilp instructions

[Sukuna0007Abhi]

Fixes #1061: Can't describe instructions with overlapping match strings

Changes

1. Fix decode tree crash (decode_tree.rb)

• Problem: inst_format.reverse[cur_range] can return nil, causing match? to crash
• Solution: Added nil safety check with range_bits && range_bits.match?
• Why needed: Without this, the decode tree crashes before it can process any overlapping patterns

2. Add missing Zicfilp instruction definitions

• Problem: Sir @ThinkOpenly identified that sspush/sspopchk were "represented with incorrect mnemonics"
• Solution: Created proper instruction definitions with constraint-based disambiguation
• Files added:
  • sspush.yaml - Variable form with not constraints (allows only x1/x5)
  • sspush.x1.yaml, sspush.x5.yaml - Fixed forms for specific registers
  • sspopchk.yaml - Variable form with not constraints (allows only x1/x5)
  • sspopchk.x1.yaml, sspopchk.x5.yaml - Fixed forms for specific registers
• Why this works: The patterns use constraint-based disambiguation where not constraints make seemingly overlapping patterns distinguishable

3. Fix Rakefile build issue

• Problem: Missing require "pathname" causing build failures
• Solution: Added the missing require statement
• Why needed: Build system needs pathname support to process the new instruction files

Ready for Review sir @ThinkOpenly @dhower-qc

[dhower-qc]

A few high-level comments:

• Good job. That decode_tree code is pretty hard to understand. I wrote it and can barely parse it ;)
• This should be two PRs; one for the decode_tree fix and one to add the instructions.
• CI fails because you haven't identified the new instructions as "hints" of mops. You'll see an example of how to do it in the mop files.

[Sukuna0007Abhi]

A few high-level comments:

• Good job. That decode_tree code is pretty hard to understand. I wrote it and can barely parse it ;)

Thanks sir @dhower-qc , Yes sir it's a pretty hard and a bit confusing but I will try my best ,

• This should be two PRs; one for the decode_tree fix and one to add the instructions.

yes sir I will create 2 PR's,

• CI fails because you haven't identified the new instructions as "hints" of mops. You'll see an example of how to do it in the mop files.

Yes sir I found it now, and also there some wrong in not constraint I think.

[Sukuna0007Abhi]

I am pointing this PR to add the instructions only

[ThinkOpenly]

I think this just converts a crash into an infinite loop. That's what I'm seeing in testing, anyway. Are you seeing something different?

[Sukuna0007Abhi]

Oops sorry sir @ThinkOpenly I think I missed to add the another needy few lines here, sorry it's because I got very sad after lfx result😞 , sure sir will fix that