Fix examples by resizing the surface

[jmjoy]

When running some examples on my Linux system, I encountered an error:

index out of bounds: the len is 0 but the index is 0

This error occurred when attempting to perform array indexing write operations on the buffer generated by surface.buffer_mut().

I successfully resolved these issues by resizing the surface during initialization.

[MarijnS95]

This directly conflicts with the idea introduced in #236 (CC @madsmtm) that creating a Surface either infers a size from the given window and/or a Resized event is always delivered somewhere in application startup before the first RedrawRequested event.

Hacking that original misplaced .resize() back in is unfortunately not going to fly. We need to instead debug what your Linux sample is doing, was this on Wayland or X11?

[MarijnS95]

More context is available in #237, too.

[MarijnS95]

Analyzing the code, it seems that the Wayland backend would have crashed if resize() is never called, because size contains None. On X11 however, without SHM support (i.e. falling back to sending all pixels over the wire), a default implementation is created with an empty Vec:

softbuffer/src/backends/x11.rs

Line 290 in fa3a918

Buffer::Wire(Vec::new())

[madsmtm]

Hmm, I'm fine with this PR? It does the same thing as the fruit.rs example does in #236, i.e. resizing the surface upon initialization (since that isn't done yet internally on all platforms).

[MarijnS95]

Then we should copy-paste that ugly edge-case, specific comment and all, to every other example?

I completely missed that we inconsistently had a weird edge-case in one but not all examples.

[madsmtm]

Then we should copy-paste that ugly edge-case, specific comment and all, to every other example?

I think so, yeah

[madsmtm]

(Or we should fix the underlying issue)

[MarijnS95]

(Or we should fix the underlying issue)

(Guess why I initially "requested changes" 😉)

[jmjoy]

Hacking that original misplaced .resize() back in is unfortunately not going to fly. We need to instead debug what your Linux sample is doing, was this on Wayland or X11?

X11

[jmjoy]

Hmm, I'm fine with this PR? It does the same thing as the fruit.rs example does in #236, i.e. resizing the surface upon initialization (since that isn't done yet internally on all platforms).

In fact, I modified it based on the example of fruit.rs.

I also think it's ugly to add such a paragraph to each example. It would be better to have one that can fix the underlying issue.