Cherry-pick inlining fix from #725

[jiribenes]

As discussed in #561 (comment), there's a missing rewrite.
This fix is necessary for the map and set examples, as well as some of my AoC solutions using them, so I'd like to upstream it, if possible.

[jiribenes]

Since the tests seem to pass, I'll just merge this. :)

[jiribenes]

Unfortunately, this doesn't seem to help with other inliner bugs I've seen:

ReferenceError: l_9 is not defined
    at out/amap.js:4182:76

It looks like the same bug as in #561:

                                            switch (v_y_166.__tag) {
                                              case 0: 
                                                return () =>
                                                  link_0(v_y_154, v_y_155, l_9, v_y_164, ks_119, (v_r_97, ks_120) =>
                                                    () =>
                                                      go_1((level_1 << (1)), v_r_97, v_y_165, ks_120, k_88));
                                              default: 
                                                switch (v_y_165.__tag) {
                                                  case 0: 
                                                    return () =>
                                                      link_0(v_y_154, v_y_155, l_9, v_y_164, ks_119, (v_r_98, ks_121) => {

opt IR

                          let cr9991647052 = run {create403246401(level867844944, v_y_1271111410644966)}
                          
                          cr9991647052 match {
                            case Tuple3335 { (v_y_126991101021747053: Map8335[A10266, Unit394], v_y_127001111031847054: List910[Tuple2249[A10266, Unit394]], v_y_127011121041947055: List910[Tuple2249[A10266, Unit394]]) => 
                              v_y_127011121041947055 match {
                                case Nil1135 { () => 
                                  let v_r_126891039510350733 = run {link8561[A10266, Unit394](v_y_1271411510744967, v_y_1271511610844968, l9890547046, v_y_126991101021747053)}
                                  
                                  go857746408(bitwiseShl225(level867844944, 1), v_r_126891039510350733, v_y_127001111031847054)
                                }
                              } else {
                                v_y_127001111031847054 match {
                                  case Nil1135 { () => 
                                    let v_r_126931079914350736 = run {link8561[A10266, Unit394](v_y_1271411510744967, v_y_1271511610844968, l9890547046, v_y_126991101021747053)}
                                    

[marvinborner]

I didn't have time to investigate this yet, but could this effekt-plots error (on master?!) be related?

https://github.com/effekt-lang/effekt-plots/actions/runs/12247898051/job/34166543555

[jiribenes]

I didn't have time to investigate this yet, but could this effekt-plots error (on master?!) be related?

https://github.com/effekt-lang/effekt-plots/actions/runs/12247898051/job/34166543555

Fascinating. I have something similar in the exact same file reported above (where the inliner fails ^)

$ effekt --no-optimize --backend=llvm day08/amap.effekt
Exception in thread "main" java.lang.RuntimeException: Cannot find block info for b_tmp_15333.
HashMap(infixEq -> Definition(List(),List(Variable(x_67,Int()), Variable(y_68,Int()))), b_k_13343 -> Definition(List(),List()), fail -> Parameter(Interface(fail,List())), b_k_11016 -> Definition(List(),List()), b_k_13296 -> Definition(List(),List()), balance -> Definition(List(),List(Variable(k_7096,Positive()), Variable(v_7097,Positive()), Variable(l_7098,Positive()), Variable(r_7099,Positive()))), bitwiseOr -> Definition(List(),List(Variable(x_232,Int()), ...

=> reported in #733 and #732 :)

[jiribenes]

On the meeting, we found out that Effekt Plots doesn't work because of this change (cc @marvinborner)