What is needed is some actual coding help and long term commitment on maintaining such themes. For years I refrained from doing needed changes to LuCI to not break existing themes and when I finally did, 3rd party stuff like "Material" or "Darkmatter" promptly broke.
If you start custom themes you need to be aware that there's a certain amount of responsibility attached to it, especially when they grow in popularity. The question is who is the lucky one to pick up the broken pieces in the end, the one having quickly cobbled together the theme a few years back, or the LuCI maintainer getting reports about it three years down the road.
From experience I noticed that I personally am not able to really maintain more then the two "officially endorsed" Bootstrap and OpenWrt themes in LuCI. Especially when aspects like responsive design come into play, the testing effort for each further theme grows exponentially.
So what am I getting here... first of all think hard if you really need a custom theme, maybe it makes more sense to invest work into an existing one to fix the small warts and make it customizable by introducing things like switchable color palettes. For example it makes very little sense to now copy the already half broken Material theme into a Material Dark one, only to end up with one further theme that needs to be fixed by someone.
Furthermore try to translate your ui and usability critiques into actionable items. Make scribbles, UI concepts, patches or photoshop mock ups and try to explain why the current state of affairs is broken and why you think that your change makes more sense.
Regarding the buttons, removing the color filling on the non primary buttons was a decision I made to implement some of the best UX button practises which you can find on sites like https://uxplanet.org/7-basic-rules-for-button-design-63dcdf5676b4 or even on the Material design spec site itself at https://material.io/design/components/buttons.html#usage (section "Types"). The intent was to de-emphasize non-primary buttons to steer user attention towards the primary page actions like "Edit item" or "Save & Apply".