resistor-color-trio: Add exercise (with extensions)

[sshine]

This adds the exercise from exercism/problem-specifications#1551.

In addition to the canonical exercise, this implementation extends:

• Add a "zero" test case (Black, Black, Black).

• Add a Data.Text hint in .meta/hints.md.

• Export both a label function and an ohms function: While the
  exercise is mainly about printing a formatted string, this nudges
  the student towards separating print and calculate.

• For totality, since Resistor allows expressing resistors in the
  megaohm and gigaohm ranges, extend the exercise so that the correct
  unit beyond kiloohms are returned.

• Extended unit tests are added, and coverage is described as such:

  Canonical tests are marked with 'x'.
  Extended tests are marked with 'y'.
  Optional (commented out) tests are marked with 'z'.

  Color       1, 2      3
  -------------------------
  Black       xyyyyy    xyy
  Brown       xyz       x
  Red         xyz       xz
  Orange      xxz       x
  Yellow      xyz       x
  Green       xyz       yz
  Blue        xyz       y
  Violet      xy        y
  Grey        xy        yz
  White       yy        y
  

  The over-representation of Black is partly caused by the fact that
  canonical and extended tests avoid edge cases like "4.7 kiloohms".
  There is some didactial point in incremental improvements, and as
  such, canonical and extended tests partially validates solutions.

  Optional (commented out) tests make the exercise considerably more
  difficult, so they are accompanied with a disclaimer that students
  may choose freely to include them or not.

  This way, students who are oblivious to the edge cases will not be
  aware of them, and students who are not (either by thinking or by
  inspecting the unit tests) will be able to choose.

[sshine]

This PR is currently marked as WIP because the example solutions don't handle the optional tests.

[sshine]

Currently no special-cases for "1 ohm", "1 kiloohm", "1 megaohm" and "1 gigaohm".

I'm unsure if they should be added as additional optional tests, or simply not be mentioned at all.

[sshine]

Also, I learned something very cool: 1.2e3 :: Int using the NumDecimals pragma.

[sshine]

I've realized that the derived Eq instance is problematic:

let ten1 = Resistor (Brown, Black, Black)
    ten2 = Resistor (Black, Brown, Brown)

ohms ten1 == ohms ten2 -- True
ten1      == ten2      -- False

Fortunately, since we only test label and ohms, we never compare Color or Resistor.

I have an idea for making a subsequent refactoring exercise where this insufficiency could come handy.

[sshine]

I've not enabled the optional tests, so CI does not reveal that the example solutions work for them.

The following demonstrates that this is the case

success-standard

$ stack test
...
label
  ...
  (Brown,Red,Red) should give 1.2 kiloohms
  (Orange,Yellow,Green) should give 3.4 megaohms
  (Green,Blue,Grey) should give 5.6 gigaohms

ohms
  ...
  (Brown,Red,Red) should give 1200
  (Orange,Yellow,Green) should give 3400000
  (Green,Blue,Grey) should give 5600000000

Finished in 0.0013 seconds
30 examples, 0 failures

success-text

The exact same.