Automotive power is filthy. It's full of big spikes that will eventually damage the little adjustable buck module at the top of your illustration.
Do those buck modules clean up power? Somewhat, but they aren't lab supplies. It depends on your operational definition of "clean". Buck converters create ripple due to how they work, but it's small and fast.
When I built a controller for an electric start diesel water pump I faced this and other problems. I used very large bypass capacitors to calm the larger spikes before putting 14.5V through buck converters for the various voltages required by servomotors and other things.
Buck converter output is good enough to run an Arduino and probably fine for an ESP32 but it was too dirty for an LM386 sound detector module like this one
Ripple coming in the power rail was amplified and detected as sound.
I cleaned it up by slightly increasing the voltage and then using a linear regulator to clamp the spikes. Things I've read suggest this won't work but it did which I find more compelling than theory. I imagine a 10nF bypass capacitor would also clean up these small fast transients but by the time I thought about that it was working and you don't fix what ain't broke.
Using a linear regulator to scrape off the ripple will also protect the ESP32 in the event that a buck converter failure passes 12V through. While the buck converter is working it will dissipate very little power clamping 6V to 5V, but if it gets 12V then it will get hot so if you design this kind of protection then you also need a heatsink and a piezo alarm to clue you in that the buck converter has failed.
How hot depends on the ESP32's power. The current will be whatever current the ESP32 requires. Suppose it needs 300mA, then 300mA must pass through the regulator with a drop of 9V5. When the motor runs, the alternator voltage is high enough to charge a car battery ie 13.8V to 14.5V.
So the regulator must dump 300mA @ 9.5V or 2.85W. Which isn't colossal, but it will get warm.
Power dissipation brings me to a question you didn't ask. Tracking devices that run continuously will run your battery down faster than you might think. Comms devices use more power than you'd expect to transmit, and you did say "hotspot". I strongly suggest your design should
- have the ESP use a MOSFET to switch power to subordinate modules like LTE
- have its own battery circuit charged by car power only when the engine runs
- monitor voltage and shuts down to avoid battery damage
You could also sleep the ESP32 and use a low power timer to wake it periodically.