Add support for eDEFAULT for bitflags (and maybe enums)

[Jake-Shadle]

Currently, we #[derive(Default)] for generated bitflags! in physx-sys, which is 0, which might not actually align with what the default is in C++.

Take PxGeometryQueryFlags

https://github.com/NVIDIA-Omniverse/PhysX/blob/85befb639eb736bc8ec7d4a2e9726ec2ab1d0434/physx/include/geometry/PxGeometryQueryFlags.h#L47-L55

Which currently generates

physx-rs/physx-sys/src/physx_generated.rs

Lines 286 to 301 in 926eee8

	/// Geometry-level query flags.
	#[derive(Debug, Clone, Copy, PartialEq, Eq)]
	#[repr(i32)]
	pub enum PxGeometryQueryFlag {
	/// Saves/restores SIMD control word for each query (safer but slower). Omit this if you took care of it yourself in your app.
	SimdGuard = 1,
	}
	bitflags::bitflags! {
	/// Flags for [`PxGeometryQueryFlag`]
	#[derive(Default)]
	#[repr(transparent)]
	pub struct PxGeometryQueryFlags: u32 {
	const SimdGuard = 1 << 0;
	}
	}

This will in most cases be harmless, particularly if the default is in fact 0, but for other cases it's either (silently) misleading or could cause crashes/incorrect behavior.

It should be relatively trivial to actually do impl Default for <> for both the enum and the bitflags if the enum in question does in fact have an eDEFAULT variant, otherwise we should leave the default off to force the user to consult the docs/code to make a decision rather than the current behavior.