New code for isosurface trace

[archmoj]

Supersedes #3311 i.e. without using gl-isosurface3d refactored code!
Totally rewritten and based on plotly.js most recent source code for mesh3d trace.
Also in this regard, the isosurface module which used to support various algorithms namely marching-cube, marching-tetrahedra & surface-nets for displaying mathematical functions is forked and improved to accept data cubes.
Using isovalue attribute the user may create one, two or multiple 3D contour surfaces.
It is possible to enable isocap surfaces too.

Various iso-surface traces could be modelled using this manual demo or this auto-update demo.

@plotly/plotly_js

[archmoj]

mikolalysenko/isosurface@master...archmoj:plotly-integration

[etpinard]

Reusing gl-mesh3d is a very nice idea. One drawback I noticed though: hover labels don't appear on the input data coordinates, they appear on the (interpolated) mesh x/y/z - which I think is ok for now, albeit not consistent with our other trace types. For example, surface hover always picks out data-coordinates (even when the surface coordinates are refined). In contrast, streamtubes hover picks the coordinates on the tubes themselves.

[etpinard]

@archmoj nice work! Your code and algo look very good, reusing gl-mesh3d makes it very easy to digest!

---

Now, I'm trying to verify concerns from #2752 (comment) - pasting

varwidth=64varheight=64vardepth=64varisomin=1600varisomax=2000varxs=[]varys=[]varzs=[]vardata=newUint16Array(width*height*depth)vark=0for(varz=0;z<depth;z++)for(vary=0;y<height;y++)for(varx=0;x<width;x++){xs.push(x/width)ys.push(y/height)zs.push(z/depth)data[k]=x<width/2 ? isomax+1 : isomin-1k++}Plotly.newPlot(gd,[{type: 'isosurface',x: xs,y: ys,z: zs,volume: data,isomin: isomin,isomax: isomax,cmin: isomin-100,cmax: isomax+100,smoothnormals: true,isocaps: true,singlemesh: true,colorscale: 'Portland',capscolorscale: 'Jet'}],{scene: {xaxis: {range: [0,1]},yaxis: {range: [0,1]},zaxis: {range: [0,1]}}})

appears to result in an infinite loop. Am I doing anything wrong?

[archmoj]

@etpinard This is a demo related to the comment that you mentioned: #2752 (comment)

[archmoj]

@etpinard demo link related to #2752 (comment)

[etpinard]

Can we close this PR, now that #3438 is up?