In Apply.semigroup test replace ListWrapper with Option

[kailuowang]

This is to address the root cause of #2648 and #2319, the target method of this test is the combine method provided by Apply.semigroup
https://github.com/typelevel/cats/blob/master/core/src/main/scala/cats/Apply.scala#L223-L225

def combine(a: F[A], b: F[A]): F[A] =
    f.map2(a, b)(sg.combine)

In the case of F being a list, since this is a product operation between the two lists, the result is a list of the size a.size * b.size
When this combine is used in the combineAllOption test, which combine all items in a Vector[ListWrapper[Int]], it exponentially increase the size of the accumulative result list. That leads to a list of millions of items and over limit GC pressure that breaks the build.

I switch it to using Option instead.

[codecov-io]

Codecov Report

Merging #2827 into master will decrease coverage by 0.29%.
The diff coverage is n/a.

@@            Coverage Diff            @@
##           master    #2827     +/-   ##
=========================================
- Coverage   94.17%   93.87%   -0.3%     
=========================================
  Files         368      368             
  Lines        6933     6955     +22     
  Branches      303      305      +2     
=========================================
  Hits         6529     6529             
- Misses        404      426     +22

Impacted Files	Coverage Δ
core/src/main/scala/cats/Apply.scala	75% <ø> (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 422c989...bfcf98c. Read the comment docs.

[rossabaker]

Looks good, but see suggestion on scaladoc.

[rossabaker]

Light copy editing (spans lines, so I couldn't GitHub it):

   * This semigroup uses a product operation to combine `F`s.
   * If the `Apply[F].product` results in larger `F` (i.e. when `F` is a `List`),
   * accumulative usage of this instance, such as `combineAll`, will result in
   * `F`s with exponentially increasing sizes.

[kailuowang]

thanks!

[DavidGregory084]

Thanks for looking into that @kailuowang!

[LukaJCB]

Thankfully we can eradicate one more source of flimsyness in the tests!