How much current are you planning on needing to measure, Vinh? Based on your recent mule pulling 16A, are you looking to max out at 20 or 30A? As someone who has needed to measure and log 10~60A DC, it is a pain.
General purpose DMMs seem to start skewing current readings much over 5A with their insertion resistance on battery powered devices that rely on low internal resistance for high current delivery. Even with custom leads. Sadly, I've never found a good de-rating spec for this.
I'd probably go the clamp on DC meter route. You can use nice stout wiring and no shunt to get an accurate measurement where internal battery resistance makes a sizable difference. A Fluke 325 with a 40A range and 10mA resolution looks like it'd work well for you. Extech has some cheaper ones: DC400 and MA220 both do 10mA resolution and they have a bunch of other models with 100mA resolution. Or if you really want to buck up, an Extech 380942 does 30A DC with 1mA resolution (
edit: 1mA resolution up to 4A, 10mA above that). Just be careful around magnetic control rings; that'll definitely skew the hall effect sensor.
Another idea is a Watt's Up Ammeter / Watt meter. RC and solar power guys like them. 0-100A with 10mA resolution at a cheap price. The shunt (seems to be a pretty standard 100A, 100mV model) should still be ok with single 18650 hot rods. 1mOhm for the shunt vs. ~40mOhm internal resistance on a Samsung 20R or ~90mOhm on a Keeppower 3400mAh.
Similarly, you could just use a 100A, 100mV shunt with a DMM.
HKJ has used cheap hall effect current sensor ICs as shown here:
http://lygte-info.dk/info/CurrentAdapter%20UK.html It's nicer for logging because you can transform the voltage data en masse with software, instead of trying to convert single DMM readings in your head/with a calculator. Various models mounted on boards are easy to find on eBay or google "robot current sensor". They tend to run in the ~1mOhm range plus board. On some boards, you could beef up the traces with solder.
Just my $0.02.