The electrical noise issue was not the enclosure, but the shielded cable between the hall sensor and the MachTach. Convention would be to ground the shield at the source (not both ends), but any grounding of the shield (either end or at both ends) made the MachTach go wild. So if you use shielded cable, my recommendation is float it on both ends. Grounding is always one of those funny things, doesn't always work the way it is suppose to.
I would go with the hall sensor, you also only need a single magnetic to KISS. The IC hall sensor provided with the kit is a bit fragile, I used it on my lathe but wouldn't want to use it again if possible. You can get the cylindrical threaded type, NPN in 8 or 12 mm diameters. You connect the hall sensor brown power to wire to +IR, the blue negative to -RCV, and the black switched sensor line to +RCV. Works very nicely and no need for a pull down resistor.
Examples of sensors:
NJK-5001C
LJ12A3-4-Z/BX
Gear tooth sensor can be a bit more temperamental, you need the assembly he provides with the two capacitors in order for it to work correctly. At least he has it as an option. One problem on mills with a back gear, is you cannot take the tach off of the drive head, you need to pull the tack signal off of the spindle so need a very compact pickup assembly. One place where the small hall sensor is more desirable. On my mill which came with a factory tach, it uses an output from the VFD for speed and they also have a back gear sensor which rescales the tach range when in back gear. I recently build two integrated Halo Light and tach sensor for some PM 949 mills that where 1/4 thick and slid up on the mill ram. They required a very small hall sensor, it was very temperamental to magnetic flux.
View attachment 256690