Buck converter inductor current calculation - Am I doing it right?

I'm looking at using this switching regulator on my board:

My output voltage will be 5V, maximum output current 3A, input voltage 5-16V. The regulator will run at 2.5mhz.

On page 19 of the datasheet it lists an equation for calculating maximum inductor current.

I am considering using this inductor because it fits the reccomended 2.2uH inductance and has a maximum and saturation current of 3A.

The problem is, when I do the calculations on page 19, I arrive at a number that seems wrong.

I tried what I think is a worst case calculation:

IL = Vout * (1-Vout/Vin) / (Lmin * Fsw)
IL = 5* (1-5/16) / (0.0000022 * 2500000)
IL = 0.625

This seems way off. I mean, I know that if I'm drawing 3A from the regulator 3A must be passing through the inductor. So why is the value I get from this calculation so low? Am I off by an order of magnitude here?

Google says a microhenrie is 10^-6 henries, which is 0.000001. Multiply that by 2.2, I get 0.0000022. So the math seems sound. What am I doing wrong?

Hm, I just found this document:

It says on the first page that the ripple current is load independent. Perhaps this means I should add 0.625 to my 3A load, and either decide that 2.3A is sufficient for my needs, or select a larger inductor capable of handing more current?

Oops, I just realized there was another equation on the previous page in the datasheet, and I'd failed to notice that one of the IL's has a little triangle in front of it and that specifies that it represents the ripple current.

Now off to the Eevblog to find out what ripple current represents. I'm sure I watched Dave give a tutorial on that recently.

Oh, and that other equation is IL = Iout + (/\IL / 2), so if /\IL is 0.625 then at 3A output IL = 3.3A, or I could have a 2.6A output with an inductor that saturates at 3A.

Hard to decide if I should go for the smaller 3A inductor:

Or this larger inductor rated for 5A:

I'm thinking I should just go for the smaller one. Costs half as much, and it actually states that it works at up to 5mhz which the Bourns does not. I assume it does, but it doesn't say, so...

I never understand why DC-DC converter datasheets try to explain
their design equations in poor english with much confusion (random
units is a good one, values in uF or kHz rather than F or Hz), when they
could simply include a 20-line C program that calculates all the
component values completely unambiguously.

This site provides good instructions on designing switch mode power supplies and allows you to simply enter
the input output requirements, and it will give you the component values.
http://www.daycounter.com/Calculators/Switching-Converter-Calculator.phtml