h5web: Prepare tomo alignement tool

This PR extend the plot layer we use in Daiquiri to prepare tomo alignment tools.

Part of the result can be checked at https://confluence.esrf.fr/display/UXUIT/TomoAlign+prototype

It contains

• All in one viscanvas (actually LinearVisCanvas2, i wll provably rename it Plot2D in the future)
  • Including view point history
  • Including ROI engine (for point, rect, vline, hline)
  • Including segments supporting dash-gap without SVG (that's the minimal i need for 99% of the use cases for now)
  • Reworked ruler

The aim is to provide in the future everything needed, dropping any SVG elements.

• Reduce the time it takes to design custom things
• Drop the discrepancy between GLSL canvas and SVG (if there is no SVG, there is no discrepency)
• Simplify PNG export (if there is a single canvas an export will be very easy)

For the future

• Rework every tomo component with that new canvas
• We could work to render text as glsl (troika-three-text and drei project)
• Merge the mouse interaction together with LinearVisCanvas2 in order to fix glitches between different mouse interactions