Update documentation of Uri.parse.

Update documentation of Uri.parse.

It currently says that parse tries to escape invalid characters.
That's not true (at least not in most cases, though I think a few might
still work).
This just removes that part of the documentation.

BUG= http://dartbug.com/23688
R=sgjesse@google.com

Committed: https://github.com/dart-lang/sdk/commit/8450179afefbfb8f88c01daa379163862c3edff5

[Lasse Reichstein Nielsen, 2015-07-08 11:27:16]

lrn@google.com changed reviewers:
+ sgjesse@google.com

[Søren Gjesse, 2015-07-09 09:05:58]

lgtm

Should we make sure it is 100% strict - or will that be a breaking change?

[Lasse Reichstein Nielsen, 2015-07-09 09:19:16]

> Should we make sure it is 100% strict - or will that be a breaking change?

That depends on what it is we allow today. My guess (without having checked) is
that we allow a lone "%", without proper hex following it, to occur and be
escaped as %25, and we allow any non-ASCII character. We probably want to keep
doing that - and not require you to UTF-8 encode first.

The thing we don't allow is any character from the "gen-delims" category being
used improperly - those are the characters used to delimit the URI parts that we
use for parsing (:, /, [, ], ?, #, @). The '[' and ']' characters are delimiters
for IPv6 addresses. It's probably harmless to have them occur in a query or
fragment, but it's simpler to not allow it and be strict with all the general
delimiters.

No plan to change anything now. If we do, we could go for being more libral in
what we accept, while still conservative in what we produce. But that does risk
someone using an invalid URL without realizing it.

[Lasse Reichstein Nielsen, 2015-07-09 10:24:31]

Committed patchset #1 (id:1) manually as
8450179afefbfb8f88c01daa379163862c3edff5 (presubmit successful).