RFC: remove the ability to use `E` rather than `e` in float literals

[bstrie]

Right now both of the following are legal:

let foo = 9.02e6;
let bar = 9.02E6;

In the interest of There's Only One Way To Do It, I don't see why having two slightly different forms for the same result is necessary. I thereby propose that the capitalized form be removed because:

1. All other letters in numeric literals (i, u, f) are lower case. Mandating e would make literals more consistent.
2. Among numbers, a lowercase letter is easier to recognize at a glance than an uppercase one, making this a readability win.

This also makes the grammar slightly simpler. But really I just care about not having two trivially different ways of doing the same thing.

[lifthrasiir]

In my opinion, this is equivalent to disallowing uppercase or lowercase hexadecimal digits (e.g. 0xABCD is allowed and 0xabcd is not). Unlike numeric suffixes there are many programs that generate both variants of float representations.

[graydon]

Yeah, I concur with @lifthrasiir , we allow mixed-case hex literals too. There are unfortunately quite a few ways to write numerically-equal numbers. This doesn't strike me as a big deal either way, I was just going with the most common grammars for FP out there. http://rosettacode.org/wiki/Literals/Floating_point

[bstrie]

Good enough for me, I'm not about to argue that we shouldn't allow mixed-case hex literals... though I'm a little tempted. :)