fix: 3816 - use a 3.2.0 mlkit version of mobile_scanner

[monsieurtanuki]

What

• Implementing @juliansteenbakker's suggestion about mlkit pod.
• Of course one day we'll have to fix it in a more decent manner, but let's try this one!

Part of

• The iOS release is blocked #3816

[monsieurtanuki]

Fingers crossed...

[monsieurtanuki]

It worked, didn't it?

[juliansteenbakker]

Looks like it worked! Ill keep the branch thats being used right now in sync with the master branch of mobile_scanner, but i'd still recommend dropping the conflicting camerawesome package so you can use the pub.dev version.

[monsieurtanuki]

camerawesome is so far only an option we're considering, as we are currently in a "relevant barcode scanners" test/benchmark phase.
It becomes less attractive if there are conflicts with the already used packages. And those conflicts are not easy to track as @M123-dev develops around camerawesome and doesn't use a mac.

[M123-dev]

Yes my guess with camera awesome was: since it's a middle thing between the bare bones flutter:camera and a fully integrated scanner it would make it simpler to use but at the same time more flexible to do things as moving the focus point (see my issue in the repo, it is possible with a bit of hacking around)

[M123-dev]

While talking about this, while we have loose discussions about if we should change the main use of the scanner the first thing we talked about is the scanner itself, but does not matter what it will get, a "full-screen" scanner would definitely look better than cutting it halfway through.

Is there a way to move the focus point up a bit in the upper two thirds of the screen in mobile_scanner @juliansteenbakker

[monsieurtanuki]

a "full-screen" scanner would definitely look better than cutting it halfway through.

That's really a matter of appreciation. As an engineer the extra work needed for making the camera work full screen with a focus on top, only to have 90% of the bottom camera output hidden by product cards, does not seem relevant.
Let alone the performance issues (more memory, more code, more processing needed) and the history of bugs (especially given the android diversity).
To me the problem is not with mlkit, zxing or camerawesome, the problem is with the carousel, cf. #3777. As a Smoothie user I never actually used the carousel (I use history instead), and I cannot name an app with a barcode/qrcode scanner where the camera/scan is not the biggest and central widget of the screen.