Fix Lazy for tagged types

[joroKr21]

Fixes #584

Funny thing, this is also a partial fix for #309, but it looks like some types are automatically dealiased by Scala (e.g. lgen.Repr and Record.{..fields}.T) and in this case scala/bug#10506 kicks in.

[codecov-io]

Codecov Report

❗ No coverage uploaded for pull request base (master@8ce40d1). Click here to learn what that means.
The diff coverage is 0%.

@@           Coverage Diff            @@
##             master    #797   +/-   ##
========================================
  Coverage          ?   87.9%           
========================================
  Files             ?      65           
  Lines             ?    1497           
  Branches          ?       5           
========================================
  Hits              ?    1316           
  Misses            ?     181           
  Partials          ?       0

Impacted Files	Coverage Δ
core/src/main/scala/shapeless/lazy.scala	0% <0%> (ø)

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 8ce40d1...fc1befc. Read the comment docs.

[melrief]

This is great, thank you @joroKr21 !

[milessabin]

Does this change actually break the test you've modified?

[joroKr21]

Yes, the unmodified test fails because of ambiguous implicits, since the workaround for #309 is not necessary when deriving a typeclass instance for case class. However, it cannot be completely removed, because it is still necessary for the tests in LabelledGeneric211Tests. The difference is that the types in LabelledGeneric211Tests are eagerly dealiased (I realized this happens in existentialAbstraction).

[milessabin]

Gotcha ... I'm going to defer this to 2.4.0 and hopefully get a fix for scala/bug#10506 into 2.13.0.

[joroKr21]

@milessabin should we merge this for 2.4.0?

[milessabin]

Yes, I think so.