[JVM] Calling .Str on &infix:<+> hits Mu::Str instead of Code::Str

[usev6]

The Problem

The test for the string value of &infix:<+> in https://github.com/Raku/roast/blob/0d696f42a5/S03-operators/names.t#L14 started to fail on the JVM backend. (I don't know when it started to fail, but it used to work at least until 2022-10.)

Expected Behavior

The output of the following command should mention the operator name -- as it does on MoarVM:

$ ./rakudo-m -e 'say &infix:<+>.Str'
Sub+{is-pure} object coerced to string (please use .gist or .raku to do that)
  in block <unit> at -e line 1
infix:<+>

Actual Behavior

Only Sub+{is-pure} is mentioned on the JVM backend:

$ ./rakudo-j -e 'say &infix:<+>.Str'
Sub+{is-pure}<1891315145>

Additional Information

If I'm not mistaken, Mu::Str is hit instead of Code::Str. Explicitly calling the latter gives the expected result:

$ ./rakudo-j -e 'say &infix:<+>.Code::Str'
Sub+{is-pure} object coerced to string (please use .gist or .raku to do that)
  in block <unit> at -e line 1
infix:<+>

Environment

• Operating system: FreeBSD 12.4-RELEASE
• Compiler version (perl6 -v or raku -v): Welcome to Rakudo™ v2022.12-1130-g5ed14bc46.
  Implementing the Raku® Programming Language v6.d.
  Built on JVM.