WIP: Refactor validation helpers into a single module

[jzaefferer]

Fixes #362

I made the functional changes, but can get past jshint:dist. Apparently the build system isn't able to handle this module type. onBuildWrite is kinda complicated, so I'm hoping for some help to debug this. Once it actually builds, we should get some numbers from the compare_size task.

I definitely like the raw reduction in LOCs, currently 291 additions and 456 deletions. Less dependencies and factory arguments is nice.

[rxaviers]

The outcome of this change is:

	Minified size	size 𝛥	Minified+gz size	size 𝛥
globalize.min.js	3906	53.18%	1672	33.23%
globalize/currency.min.js	5751	-3.68%	2469	-5.44%
globalize/date.min.js	15243	0.07%	4959	-0.74%
globalize/message.min.js	15892	-3.60%	4920	-4.84%
globalize/number.min.js	7226	-0.17%	2965	-1.66%
globalize/plural.min.js	5270	-2.68%	2231	-1.76%
globalize/relative-time.min.js	1468	0.69%	697	-1.55%

The whole purpose of keeping each validation separately is being able to balance the weight, i.e., only including them when necessary. In other words, currently no unnecessary validation is loaded. After the change, it is.

Having said that, I'm mostly concerned/interested in seem the delta difference in the runtime modules (#398). Because, they use even less validations, therefore the impact will be bigger.

[jzaefferer]

What do these numbers mean?

[rxaviers]

The size of final modules.

[jzaefferer]

Specifically, how can I read the percentages?

[rxaviers]

Delta sizes (comparing master with this PR)

[jzaefferer]

So the main module gets bigger, the others get smaller?

Doing the comparison on top of the runtime changes seems reasonable.

[rxaviers]

So the main module gets bigger, the others get smaller?

yeap

[rxaviers]

I am closing all old (and stalled) PRs