Designing a Microcontroller-Based Electronic Fuse for PWM-Controlled 10A Incandescent Loads

I'm working on a project now that involves a microcontroller driving some high DC currents, an example of the problem I'm facing is that one of the outputs is a high side mosfet driven 10 amp output.  The load is undefined, but will not be a motor, or anything really tricky like that.  It will in many cases though, be an incandescent bulb, the running current of a typical bulb is around 5-6amps, but the cold current is much higher than the 10 amp limit, hard to measure it as it's transient, but based on what it trips, it's somewhere between 12-15amps.  The main purpose of the device is to prevent bad stuff from happening due to things like electrical shorts, or bad installations pulling too much current from the device.  To throw another wrench in the monkeyworks, it is desirable to be able to pwm (60-100hz or so) the output for variable brightness.  Anyone interested in touching this one?  My current paths of thought are as follows: 
1- inrush current limit controller such as zxct1032, however, when the inrush cap is set high enough to stay on long enough for a bulb to start up, it's on too long to prevent excessive fet heating during a dead short.  2- ina181 type current amplifier, running the output to the MCU ADC, have software monitor current and turn off the output if over current limit, this would allow momentary overcurrent during startup then intelligently monitor current while running.  downsides may be processing overhead and possible software bugs causing catastrophic failure
3- current sense amplifier/comparator chip, with discrete components determining the fault current, the flag of which would run to the MCU and trigger a pin change interrupt, which could then determine whether the overcurrent was within the startup time and then intelligently switch off the fet or leave it on until the bulb heated up.  downsides would be more discrete parts, upsides might be more reliable software with less overhead (interrupts) ...

it's a cool project, just wondering if anyone else has run into something similar and might have an elegant solution

Indeed the project is cool and I'd love to involve in.

You may want to look at a device like the IR3313. You can program the trip current via a resistor and even measure the current flowing by measuring this feedback voltage, although it will need calibration. I have found this device to be incredibly robust and we have used it for many applications that sound similar to yours. I seem to remember though, that its switching speed was quite slow,  although it may work up to 60Hz PWM. Looking at the rise and fall times, suggests it is capable of +2KHz so maybe I am mis-remembering my target PWM.
There are quite a few devices like this. Infineon has some more, ON semi does and ST as well. On a project that required a higher PWM frequency (in the kHz) I used the VN751PT.  But it doesn't have the programmability of the IR3313.