I use data published by JPL/NASA HORIZONS interface which provides orbit data for all natural bodies and spacecraft. The "glitch" is because of the data I fetched and used - it has nothing to do with the animation per se.

As of now, the JPL site provides data only until 20th August - the time when I fetched the data a few days back, the data was available until 16th September. This was a "merge" of 3 orbit segments - the Earth bound orbits, the transfer orbit and lunar bound orbits. The former two fit the observed data. The latter was entirely based on predictions. This merged orbit has a discontinuity where the predictions start and that's why there seems to be a glitch.

Once revised data is available from JPL/HORIZONS, I presume the glitch will go away.

Not sure! Author of this tool is here: https://forum.nasaspaceflight.com/index.php?topic=20324.msg1...

He might visit HN if notices traffic from here :)

One pretty accurate way of approximating orbits is the patched conic method where you break your n-body problem into multiple 2-body problems picking the most influential body for specific parts of a journey.

Interplanetary journeys will have three parts ex. earth-craft, sun-craft, and mars-craft.

The glitch might be some noise switching between two different patches.