[FEAT] Add bilinear resize to RawImage

[BritishWerewolf]

I wanted to make a start implementing various resizing algorithms for RawImage.
Here is an example of the original canvas API implementation and this PR's bilinear implementation.

I have performed the same operation using both implementations, and cropped the same section to demonstrate the level of detail making it slightly easier to review.

Below is the original image.

And this is the code that I used to generate these images:

const resamplingMethods = ['bilinear', 'canvas'];
const sizes = [0.25, 0.5, 2];

for (let resample of resamplingMethods) {
    for (let size of sizes) {
        image
            .resize(Math.round(image.width * size), null, { resample })
            .then(image => image.save(`${resample}-${size}x.png`));
    }
}

Implementation	Resize	Image	Crop
Canvas	0.25x
Bilinear	0.25x
Canvas	0.5x
Bilinear	0.5x
Canvas	2x
Bilinear	2x

---

Additionally, I realised that it is possible that the width or height could technically be a float, and so I've added an exception if the user passes in a float, and just rounded the automatic value that is assigned if the "maintain aspect ratio" route is taken.

[BritishWerewolf]

The implementation of the algorithm could definitely be improved, because the Canvas API tends to have better results, however a bad implementation is better than none.

If you are not happy with this, I am more than happy to improve it, but my current plans are to go through the current supported resampling methods and implement those in some respects - then improve them going forward.

[BritishWerewolf]

I'm not sure if my comment makes sense, but basically if we had an area like this:

array
1 2 3 4 5 6 7 8 9 10 11 12

image
1  2  3  4
5  6  7  8
9 10 11 12

To get the element on the third row (3 - 1 since it is zero indexed) and second column (2 - 1), we must multiply by the width (4), then add the column.
In this case, we would do:

((row - 1) * width) + (column - 1) = index
((3 - 1) * 4) + (2 - 1) = 9

[BritishWerewolf]

Here, we need to get the x value for the top and bottom of the new resized area, then interpolate between these two values to find what the new y value.

topLeft          topRight
+--------T--------------+
|                       |
|                       |
|          I            |
|                       |
|                       |
+------------B----------+
bottomLeft    bottomRight

Here, top is represented by T, bottom is represented by B, and interpolatedValue is represented by I.