Have implementation for --no-* flags reflect documentation.

[indexzero]

The current documentation does not reflect the actual implementation. Quoting from that documentation

// flag with no value (-f, --force)
force: flags.boolean({
  char: 'f',
  // by default boolean flags may also be reversed with `--no-` (in this case: `--no-force`)
  // the flag will be set to false if reversed
  // set this to false to disable this functionality
  // allowNo: false,
}),

In cases such as this there can only be two outcomes: the docs are wrong, or the implementation is wrong. This PR assumes the implementation is wrong and updates the behavior to reflect the docs.

Remark: the documentation itself is somewhat ambiguous, but "set this to false to disable this functionality" implies that it is enabled by default.

[codecov]

Codecov Report

Merging #33 into master will increase coverage by 0.33%.
The diff coverage is n/a.

@@            Coverage Diff             @@
##           master      #33      +/-   ##
==========================================
+ Coverage   90.39%   90.72%   +0.33%     
==========================================
  Files          11       11              
  Lines         302      302              
  Branches       80       80              
==========================================
+ Hits          273      274       +1     
  Misses         10       10              
+ Partials       19       18       -1

Impacted Files	Coverage Δ
src/flags.ts	100% <ø> (ø)	⬆️
src/parse.ts	91.97% <0%> (+0.72%)	⬆️

---

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 d2dac2d...4d37d92. Read the comment docs.

[jdx]

I feel we should change the docs actually. I prefer that implementation and this is a breaking change

[indexzero]

@jdxcode I don't have strong feelings one way or the other. Requested documentation update in updated in oclif/oclif.github.io#13