Add new shapes to GaeaNodeFalloff

[Zehir]

Rework falloff node with new shapes

Description

This PR is a rework of the falloff node with some new features and bug fix.

Because an image is still better then a lot of words, here the old node and the new node. In the screenshot beflow the FalloffMap2 is the new version of the FalloffMap.

There is now 4 shapes included with the node. I did not include each variation in the add node tree, just the basic FalloffMap node like before.

• Square, like before just a square
• Rounded Square, a square but with round corner
• Circle, just a circle
• Squircle, a circle in the middle but a rounded square on the border.

Sample images

Works also with lower "resolution", the old one had an offset issue that is now fixed;

It's also support non square shapes (I used the PR #366 for the preview);

Q&A

• Why there is that many var defined in the FalloffSampler class ?
• It's because of performances. It's some calculation that need to be do only once and not for each cell.

Performance tests

I tried to generate a big grid and it's taking in average 530 ms to generate the data.
I tried without all var it's taking 550 ms in average
I also tried with a PackedFloat32Array but was worse: 580 ms

For a standard grid of 64*64 it's taking 4 ms

References

https://old.reddit.com/r/proceduralgeneration/comments/kaen7h/new_video_on_procedural_island_noise_generation/gfjmgen/
https://www.redblobgames.com/maps/terrain-from-noise/#islands

[BenjaTK]

The squircle should probably be removed, rounded square does almost entirely the same

[Zehir]

The squircle should probably be removed, rounded square does almost entirely the same

I would prefer keep it because it's not exactly the same and it's just 3 more lines of code

[cullumi]

The squircle should probably be removed, rounded square does almost entirely the same

I would prefer keep it because it's not exactly the same and it's just 3 more lines of code

There's probably no real downsides to holding onto, yeah. While the differences are subtle, they are technically different.

[BenjaTK]

Missing hints for start and end to be withing 0.0-1.0.

[Zehir]

Missing hints for start and end to be withing 0.0-1.0.

They should not be clamped to 0.0-1.0, you may want use values outside the "standard" range

[BenjaTK]

LGTM