Discrepancies in engine evals

I was analyzing a game from a book I'm reading at the moment call The Art of Attack in Chess by Vladimir Vukovic. It's a crazy game in the Vienna game from 1872 between Hamppe and Meitner. The game ended in a draw after an incredible King march (I'll post the game below), but after 10. KxQ, the engine is convinced White is completely winning. It's only after playing through all the moves that it changes from like +10 to 0.00! Ok, I get that engines can't handle these weird positions sometimes, but is it a bug that the evaluation bar will show one evaluation and the notation window another as in the following screenshot?

For those interested in this incredible game, here it is:

indeed

When it comes to analysis, I think it's fine to use them.