Fix swap.cow.fi appData bug

[W3stside]

Summary

Fixes issue raised by @nlordell here

Regex test:

[github-actions]

• 🔭 GP Swap: CoW Protocol v2 Swap UI

[nlordell]

🎉

Just one concern about additional spaces in the regexp.

[anxolin]

Great! although I would think that the problem was also that we are sending 0x0 if there's not a known pattern in the URL

I would add some metadata as the default, so we differenciate between us sending orders from "known" domains and others sending orders with all 0s

[github-actions]

CLA Assistant Lite All Contributors have signed the CLA.

[anxolin]

I just had a nice idea for using as a default.

Because @alfetopito is making the meta-info dynamic, maybe we will want to use always the domain in the appCode field of the meta-information.

This way, this info will be already part of the meta-info, making it more informative. We don't need to do patter matching any more to decide the appData, its always dynamic.
So now, on our upcoming release of the "slippage tolerance" we can potentially change this easily.
Nothing to address in this PR

WDYT?

[anxolin]

I think the underlying issue was not solved as explained here #698 (review)

However, I think is not worth doing it, if we do what I suggest here:
#698 (comment)

[alfetopito]

I just had a nice idea for using as a default.

Because @alfetopito is making the meta-info dynamic, maybe we will want to use always the domain in the appCode field of the meta-information.

This way, this info will be already part of the meta-info, making it more informative. We don't need to do patter matching any more to decide the appData, its always dynamic. So now, on our upcoming release of the "slippage tolerance" we can potentially change this easily. Nothing to address in this PR

WDYT?

In my PR, I still rely on these mappings as fallback, and also use that to set environment key

[W3stside]

@anxolin maybe make an issue for your concern, @alfetopito can coordinate as I'm not familiar with that code so much. merging this

[nlordell]

I would add some metadata as the default, so we differenciate between us sending orders from "known" domains and others sending orders with all 0s

The advantage of making it all 0s is that it is easier to notice issues. If the app_data had been some random bytes, I probably wouldn't have noticed that we were sending "unknown domain" app data.

I agree that the "holy grail" would be to make it entirely dynamic and compute some app data based on the current domain.