My organization has developed a custom tiling scheme, because our layers cover a fairly small area and our users complained there weren't enough scale levels in the areas they wanted to see. Since the custom scheme increments by less than double the previous value, we can't use it for vector tile layers.
Why do the levels need to double for vector tile layers, but not raster based caches? Is it for technical reasons, or optimization? Is there any way to override this?
Unfortunately, no. From the document you linked, in the syntax table:
"Tiling scheme from an existing vector tile service will be used. Only tiling schemes with scales that double in progression through levels and have 512-by-512 tile size are supported."
Regardless of whether you use an index or flat tile generation, it wants the scales to double.
Here is what our Documentation says about applying a custom tiling scheme.
"A vector tile service or tiling scheme file to be used if the service_type parameter is set to EXISTING. The tiling scheme tile size must be 512 by 512 and must have consecutive scales in a ratio of two." - https://pro.arcgis.com/en/pro-app/tool-reference/data-management/create-vector-tile-package.htm
For vector tiles it's very important that the tiles have this 1:2 ratio. This ensures the levels of detail both Overlap and fit within the Root tile, (LOD 0 in this example).
LOD 0 @ 5.91657527591555E8
LOD 1 @ 2.95828763795777E8
LOD 2 @ 1.47914381897889E8
Note the Scales here are taken from the AGOL tiling scheme. The scale values are halved for each LOD, resulting in 4 tiles from every 1 tile of the preceding LOD.
Hope this visual helps.
Esri Support Services.