Add DTLS-in-STUN (SPED) attributes

[fippo]

defined in pion/stun#260

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 88.39%. Comparing base (91a3151) to head (2a77adf).
⚠️ Report is 1 commits behind head on master.

Additional details and impacted files

@@            Coverage Diff             @@
##           master     #861      +/-   ##
==========================================
- Coverage   88.47%   88.39%   -0.09%     
==========================================
  Files          43       44       +1     
  Lines        5512     5540      +28     
==========================================
+ Hits         4877     4897      +20     
- Misses        441      446       +5     
- Partials      194      197       +3     

Flag	Coverage Δ
go	88.39% <100.00%> (-0.09%)	⬇️

Flags with carried forward coverage won't be shown. Click here to find out more.

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[fippo]

I don't think we have consensus on the maximum length to put in the spec but this matches what is currently in libwebrtc.

[JoTurk]

Hello, All of our stun attributes serialization/deserialization helpers live in stun, why this live in ice and not stun? i think if we move them to stun we can merge that right away, unless there is something i'm missing.

[fippo]

We have priority.go (priority), icecontrol.go (controlled/controlling) and usecandidate.go (use-candidate) in this repo though.
Which makes sense, ICE is an extension of STUN and these are just the attributes mentioned in https://www.rfc-editor.org/rfc/rfc5245#section-19. sped is similar in that respect.

[JoTurk]

We have priority.go (priority), icecontrol.go (controlled/controlling) and usecandidate.go (use-candidate) in this repo though.

All of these examples are from the library transition d895187#diff-034522ab4c2a896e9b3be8605c36afbb252cbc0648ff9f1378227a484ab34fda I think the reason they were added here was because of the STUN library transition, and this was to make it possible to swap from gortc/stun to pion/stun attributes without a breaking change.
But we also added nomination attribute few month ago which i kinda regret. This is the reason for my comment.

ice/renomination.go

Lines 71 to 86 in 91a3151

	func Nomination(value uint32) NominationAttribute {
	return NominationAttribute{Value: value}
	}
	// NominationSetter is a STUN setter for nomination attribute with configurable type.
	type NominationSetter struct {
	Value uint32
	AttrType stun.AttrType
	}
	// AddTo adds a Nomination attribute to a STUN message using the configured attribute type.
	func (n NominationSetter) AddTo(m *stun.Message) error {
	attr := NominationAttribute{Value: n.Value}
	return attr.AddToWithType(m, n.AttrType)
	}

The way i see it, is that that the user shouldn't not add another dependency to use attributes helpers, and we should have them in one place in case we wanted to change the interface (like we did to all of our rtp extensions few weeks ago pion/rtp#344)

but i guess we can change this later in the next ICE major release, I'll added it to the issue.
Go ahead with the merge.