Why not a clamp meter like the UNI-T UT210E? Then you can measure current in-circuit without any need to disconnect wires. I know you said you don't like hall-effect sensors, but many of us have tried the UNI-T and it works fine.
I have the UT210. Maybe I got a lemon, but I'm not getting anywhere near the precision that I've seen other reviewers get, and it's too erratic.
Under perfect conditions, with a clean supply and constant load, it's "ok", but too many things interfere with it. Noise on the line gets confused readings, it seems to latch sometimes on a particular value when I know that I'm adjusting the load and therefor should be seeing a change, and the 2A range sometimes just decides to go down to reading 0, mid-test, for no apparent reason. Also by applying very slight pressure on the clamp with my thumb and forefinger I can easily get the reading to fluctuate by 5% (and then I'm left wondering what amount of pressure on the clamp gets me the more accurate measurement).
I'll use it when I have no way to splice a connection, but otherwise I don't find it reliable.
Also, there's no theoretical difference between wiring the Metrahit Energy into a circuit and wiring two multimeters into a circuit. Both solutions will measure current and voltage simultaneously. The Metrahit can calculate power, but you didn't say you needed to do that.
Still, it's hard to beat the combination of features -- one device for voltage and current, plus a wattage reading (which is convenient, even if I don't need it all the time), plus automatic range switching between the shunts. It's just an elegant solution for a very high price.
Some 4 digit multimeters have 1 milliamp resolution on their 10+ amp range; isn't that enough? I almost always use the high current range in circuits where the current is not limited by impedance.
Yeah, that's what I'm using now, but for some applications the convenient range is 0.1mA to 1A. LED driving circuits for example. Also any circuit where I need to see what happens when the load shuts off and there's some mechanism that puts a DC-DC controller into sleep mode (they don't always go to their quiescent current immediately, sometimes it takes a few seconds).
I realize this isn't a common requirement, it would just be very convenient to have.
Gossen and a few other brands only have one current input terminal and can handle from uA to A without moving any wires or breaking the circuit.
But you always have to watch out for the burden voltage, it can be fairly bad in the high mA range.
Another solution is a power supply with a good current measurement. I use Keithley 2280S-32-6, it can measure below mA to its maximum current at 6A, always maintaining a constant output voltage.
I do a lot of tests with batteries so I have to interrupt the circuit to test it, but yeah, for everything else I'm going to have to invest in a power supply with precise readings at some point.
Take a look at the Mooshimeter: "Measure from -10A to 10A in 2uA steps".
It needs another device as a display, and introduces problems with firmware updates and wireless connectivity issues. I think it might cause more problems that it would solve, but I'll look over reviews and see what others say.
Maybe I should have prefaced the question with "I can't spend more than around $200 right now". There are some great solutions, but they're mostly very expensive.
I've been thinking of rolling my own with an INA138 and some precise ADC and voltage reference (an LM4040B?). Maybe just start with a test circuit with the INA138 and an arduino so I can get an idea of what I can expect, and gradually add precision parts to see what the added value is.