The Tesla BMS has no ability whatsoever to shunt current around bricks during charging.
The only control lever it can pull at the cell group level is to enable or disable 100mA of bleed resistors. They can only discharge groups. There's zero ability to charge them or bypass them or anything like that.
In fact, I'd have some serious doubts about this colleague's experience, at least with large EV packs, as no BMS that can bypass bricks during charging exists to the best of my knowledge. That'd require pretty massive hardware for every cell group and be insanely complex and expensive. You'd basically need contactors on every cell group.
Electrically you can't "bypass" a group with a resistor. This makes no sense. If you place a resistor across a group you're just discharging that group through the resistor, not bypassing anything.
No. If you're charging at say, 40A/240V/~10kW, you're getting ~25A DC at each cell group (current is constant in series). On the cell groups with balancers you're able to take 0.1A of that and bleed it as heat, so those get 24.9A instead of 25A. Basically a rounding error. Sure, a theoretical BMS could use more resistors to convert more incoming charge to heat... even 25A worth of resistors to make one group not charge at all by dissipating ~100W of heat from that group...... but no BMS does this. I could see something in say, an RC hobby lithium pack having bleed resistors that could offset a slow charge of an amp or so, but nothing like this exists in the EV space.
Since 100mA is peanuts in the face of 200,000 to 300,000 mAh cell groups, it's completely necessary for the Tesla BMS to enable them even when not charging to correct some imbalances. It'd take these bleed resistors nearly a week to correct a 2% imbalance! It takes about 9 months to fully discharge a pack starting at 100% charge using just the bleed resistors. (I've done this with packs that are simply unsafe to work on. Tesla even has an official procedure for this.)
Fortunately, Tesla builds the packs with very well matched original cells, so balancing is almost not even needed aside the occasional nudge to the higher cell groups. Generally deltas of less than 10mV in healthy packs. This is why hack jobs like module replacements and shops on YouTube claiming to "repair" modules by doing insane things like clipping cell fuses are 110% full of crap. The Tesla BMS has no ability to cope with modules that don't perfectly match up with the rest of the pack... as in, they need to be built and used together from day 1.
The bleed resistors serve several other purposes. Tesla's BMS uses these in clever ways to help measure cell group internal resistance, and even capacities on a per-group level, using a somewhat complex algo, down to about +/- 250 mAh... which is insanely impressive. They can use more clever methods to measure the sense line resistances, and almost immediately detect any event which causes a single cell fuse failure (as long as the BMS is active at the time.... if a shop that doesn't have a clue what they're doing snips a cell fuse and reinstalls, it'll take the BMS some time to realize it before your car's a paperweight).
Tesla's BMS is quite impressive. There's nothing like it on the market.