Refactor lngPathCorrector

[capellini]

Refactor lngPathCorrector to have the following signature:

lngPathCorrector(
  { defaultLanguage: string, allLanguages: string[] },
  { href: object | string, as?: string },
  language: string
): { as: string, href: object }

Note that, to provide the greatest flexibility, I assumed that the href object need not contain a query key, if passed in by the user. I also do some type checking to ensure that href is either a string or an object and as, if defined, is a string.

Also, in my testing, I noticed that, should a parsed URL object that contain both a search and query key, the search key string overrides any changes made to the query object.

> const { format, parse } = require('url')
undefined

> const foo = parse('/foo?option1=value1#hash1', true)
undefined

> foo.query.lng = 'de'
'de'

> foo
Url {
  protocol: null,
  slashes: null,
  auth: null,
  host: null,
  port: null,
  hostname: null,
  hash: '#hash1',
  search: '?option1=value1',
  query: { option1: 'value1', lng: 'de' },
  pathname: '/foo',
  path: '/foo?option1=value1',
  href: '/foo?option1=value1#hash1' }

> format(foo)
'/foo?option1=value1#hash1'

> delete foo.search
true

> format(foo)
'/foo?option1=value1&lng=de#hash1'

Considering in past PRs we had such trouble with manipulating a string query, I opted to remove the search key from the href object, should it exist (for example, if a user passes in a URL object as the href object) and, if necessary, add a query object to href, should it not exist (for example, if a user passes in an object without the query object in it). I thought that this was a better option than attempting to add lng to both the query object and the search string.

Resolves #155.

[isaachinman]

@capellini Sorry, just getting back to this. Can you resolve the conflict?

[capellini]

@isaachinman No worries. Rebased.

[isaachinman]

@capellini This looks good, not sure I have any technical feedback, actually. Did you do any manual regression testing?

[capellini]

Did you do any manual regression testing?

Yes, I did, but if you'd prefer, we can wait on this until #186 is merged, which should give us some added confidence that it works.

[isaachinman]

I merged #186. Can you rebase here and force push to rerun the CI? Then we can merge this.

[capellini]

Rebased.