I like this, if only for the much easier model of understanding. But I'd like to note two points:

The account doesn't visit again until next Saturday, and has now wasted any potential regeneration cycles that could have been occurring Tuesday through Saturday.

Then when visiting a week later, would be able to do the same again, allowing them the full benefit of their SP/Vest balance.

I don't think this is a characteristic of the model, only an effect of the chosen parameters. The reason you didn't "waste" two regeneration days is because you picked 7 days to recharge to full instead of the current five. The major difference, of course, is that "tokens" count as full power instead of the proportional weight votes have today.

In fact, apart from the easy cognitive model of tokens, the same result would be reached right now if the only change was that the votes always weighted the same regardless of power spent, wouldn't it?

Second, I don't think this model helps with the bot issue, in fact it might help them allowing them to vote with full power. There's no more timing calculations to get full weight, sure, but every time you don't use all your tokens the bot is ahead. At minimum they are in the same footing as human, which I believe is something they want to prevent.

Finally, I can't help to think that mostly bots would benefit from being able to buy votes when calculating that potential curation reward is higher than the cost of a token. Still, intriguing idea and worthy of further thought.