crash in `CpuWriteGpuReadBuffer::push`

[emilk]

On v0.4.0 we see analytics crashes on:

re_renderer/src/allocator/cpu_write_gpu_read_belt.rs:80

Crashing code:

    pub fn push(&mut self, element: T) {
        self.as_slice()[..std::mem::size_of::<T>()].copy_from_slice(bytemuck::bytes_of(&element));
        self.unwritten_element_range.start += 1;
    }

Callstack:

  14: re_renderer::point_cloud_builder::PointCloudBatchBuilder<PerPointUserData>::extend_defaults
  15: re_viewer::ui::view_spatial::scene::scene_part::points3d::Points3DPart::process_entity_view
  16: <re_viewer::ui::view_spatial::scene::scene_part::points3d::Points3DPart as re_viewer::ui::view_spatial::scene::scene_part::ScenePart>::load
  17: re_viewer::ui::space_view::SpaceView::scene_ui
  18: <re_viewer::ui::viewport::TabViewer as egui_dock::TabViewer>::ui
  19: egui::containers::frame::Frame::show_dyn
  20: core::ops::function::FnOnce::call_once{{vtable.shim}}
  21: egui::ui::Ui::scope_dyn
  22: egui_dock::DockArea<Tab>::show_inside
  23: core::ops::function::FnOnce::call_once{{vtable.shim}}
  24: egui::ui::Ui::scope_dyn
  25: re_viewer::ui::viewport::Viewport::viewport_ui
  26: core::ops::function::FnOnce::call_once{{vtable.shim}}
  27: egui::containers::frame::Frame::show_dyn
  28: egui::containers::panel::CentralPanel::show_inside_dyn
  29: core::ops::function::FnOnce::call_once{{vtable.shim}}
  30: core::ops::function::FnOnce::call_once{{vtable.shim}}
  31: egui::containers::frame::Frame::show_dyn
  32: egui::containers::panel::CentralPanel::show_inside_dyn
  33: re_viewer::app::AppState::show
  34: core::ops::function::FnOnce::call_once{{vtable.shim}}
  35: core::ops::function::FnOnce::call_once{{vtable.shim}}
  36: egui::containers::frame::Frame::show_dyn
  37: egui::containers::panel::CentralPanel::show_inside_dyn
  38: egui::containers::panel::CentralPanel::show_dyn
  39: <re_viewer::app::App as eframe::epi::App>::update
  40: egui::context::Context::run
  41: eframe::native::epi_integration::EpiIntegration::update
  42: <eframe::native::run::wgpu_integration::WgpuWinitApp as eframe::native::run::WinitApp>::paint
  43: eframe::native::run::run_and_return::{{closure}}
  44: winit::platform_impl::platform::x11::EventLoop<T>::run_return::single_iteration
  45: winit::platform_impl::platform::x11::EventLoop<T>::run_return
  46: eframe::native::run::run_and_return
  47: std::thread::local::LocalKey<T>::with
  48: eframe::native::run::wgpu_integration::run_wgpu
  49: eframe::run_native

[Wumpf]

found the cause: Happens when exceeding the max number of points, making this a regression (we had this bug in the past and reintroduced it)

Related to #957 (lift max number of points restriction). But short-term we can fix this more easily of course, worth thinking some more about how we're going to fix that that point limit though