Update has() to cache results and not repeat tests

[bnjmnt4n]

I've switched has to an IIFE returning an object with support info. This prevents tests from being repeated as well, ie.

if (!has("json")) {
  if (!has("json-stringify")) {
    // ...
  }
  if (!has("json-parse")) {
    // ...
  }
}

repeats the json stringify and parse support tests twice each.

[ghost]

I used the has("...") form for compatibility with has.js-aware optimizers, like RequireJS. The optimization tool will replace the has call with a Boolean primitive, and unnecessary branches (along with the has function) can then be removed during minification. I think this pull request would break that behavior, though—this is why the has function is self-contained. +@unscriptable @jaredcacurak

I could store the test results as properties on the has function itself, though—that shouldn't interfere with Uglify's or Closure Compiler's dead code removal. I remember that the has function wouldn't be removed if I stored the test results in closed-over outer variables; this may no longer be the case. I'll have a look.

[bnjmnt4n]

Thanks anyway, I might just rewrite the tests to cache the results on has when I'm free.

[unscriptable]

Yah, I don't like that the has(id) syntax requires strings in the code. Seems so primitive. Anybody want to write a has2.js? ;)

[bnjmnt4n]

Will close this for now and open new PR

[bnjmnt4n]

Now at #41