fix: revert change that broke loading framework packages

[IgorEisberg]

The changes made in ResAttrDecoder.java were reverted as they break loading framework packages during decompilation.
This fixes #3181.

Added minor redundancy cleaning as a bonus.

[iBotPeaches]

Okay thanks for the find. This looks like it reverts a regression, but also reverts a fix. So I want to use your baseboard of a find to dig into what happened here of chain loading the packages from presumed wrong pkgIds.

[iBotPeaches]

Okay I did some peeking and I must have been drinking on the last patch I did to that area. I'm lost as to why I included attribute resolution for resIds for the VALUE.

This code even prior to my fix was calling out to the resource table constantly - it just was shifting and lying about its resourceId to the current package, thus avoiding that rapid call out that you noticed when I removed it.

There must be a situation (so I gotta find an old ticket) - where obfuscation comes into play and we have to chase down a value back to the resource pool. So while you are presently today getting hammered with calls to framework pulls - it was always happening prior just being lied/skipped to current pkgId.

I am working on a long-term fix to properly adjust the parser without breaking private namespaces.

[iBotPeaches]

Closing as I merged #3187, which I believe resolves this. The other changes are unrelated and appear to be cs related