Data in the analog view is rounded to 0 decimal places in the "number under cursor" view. For examaple, a voltage that visibly fluctuates in the plot between 3.1 and 3.5 V shows as "[3 V] 2 V/div" no matter where I place the cursor.

For comparison, I've patched a locally checked out version to pass 2 instead of 0 to the `format_value_si` invocation in `pv/views/trace/analogsignal.cpp`, and now I get usable readings. (Other readings now say 745.12mV instead of 745mV).

I propose two ways to proceed with this:

* Just change the 0 to a 2.

  I can commit that to a fetchable location if you prefer, but you're probably faster just editing it.

  That variation would be good enough for my purposes, and is what I recommend.

* Change the semantics of format_value_si's precision argument from "digits after the decimal point in whichever unit was chosen" to "number of non-zero digits". (No other code in pulseview appears to use that function).

  That variation would be more correct from a significant-digits point of view (but we don't have that information to tell how many are significant, and'd need to guess). I'd offer to write the few lines of log10-round-and-clip code if that's what you prefer.