rotary-set seems busted

Hi, I am working on a C pi lib for [this quad-encoder seesaw thing](https://www.adafruit.com/product/5752), and have been trying to reverse-engineer from working python-code, but I noticed an error with setting a rotary-value:

```py
import board
import busio
from adafruit_debug_i2c import DebugI2C

import adafruit_seesaw.seesaw

i2c = DebugI2C(busio.I2C(board.SCL, board.SDA))
seesaw = adafruit_seesaw.seesaw.Seesaw(i2c, 0x49)

import adafruit_seesaw.rotaryio
encoders = [adafruit_seesaw.rotaryio.IncrementalEncoder(seesaw, n) for n in range(4)]

encoders[0].position = 100
encoders[1].position = 200
encoders[2].position = 300
encoders[3].position = 400
```

Results in this, as I would expect:

```
get version and stuff:
I2CWRITE @ 0x49 ::
I2CWRITE @ 0x49 :: 0x0, 0x7f, 0xff
I2CWRITE @ 0x49 :: 0x0, 0x1
I2CREAD @ 0x49 :: 0x87
I2CWRITE @ 0x49 :: 0x0, 0x2
I2CREAD @ 0x49 :: 0x16, 0x78, 0xdb, 0x17

set values
I2CWRITE @ 0x49 :: 0x11, 0x30, 0x0, 0x0, 0x0, 0x64
I2CWRITE @ 0x49 :: 0x11, 0x31, 0x0, 0x0, 0x0, 0xc8
I2CWRITE @ 0x49 :: 0x11, 0x32, 0x0, 0x0, 0x1, 0x2c
I2CWRITE @ 0x49 :: 0x11, 0x33, 0x0, 0x0, 0x1, 0x90
```

bytes seem to be reversed int32's, for example `0x2c, 0x1, 0, 0` is `300`.

When I read the same rotary, I get all 0's though, so I don't think it worked:

```py
print(f"{encoders[0].position} {encoders[1].position} {encoders[2].position} {encoders[3].position}")
```

```
I2CWRITE @ 0x49 :: 0x11, 0x30
I2CREAD @ 0x49 :: 0x0, 0x0, 0x0, 0x0
I2CWRITE @ 0x49 :: 0x11, 0x31
I2CREAD @ 0x49 :: 0x0, 0x0, 0x0, 0x0
I2CWRITE @ 0x49 :: 0x11, 0x32
I2CREAD @ 0x49 :: 0x0, 0x0, 0x0, 0x0
I2CWRITE @ 0x49 :: 0x11, 0x33
I2CREAD @ 0x49 :: 0x0, 0x0, 0x0, 0x
```

Thinking it might be swapped, I tried this too, but still got 0:

```py
i2c.writeto(0x49, bytes([0x11, 0x32, 0x2c, 0x1, 0, 0 ]))
```

I also tried waiting for a bit, and looping, and it stays 0, so I don't think it's a timing issue (reading too fast after setting.)

Could it be bunk firmware, or is it a lib issue?

Also, I'd like to say great & useful contributions ya'll are making to circuitpython/blinka. It's very helpful to essentially be able to use the same easy-to-use python libs on any device I might work with, and I would definitely be using that, if I didn't require C (working on native puredata extensions, for use on pi, for a few devices.) `DebugI2C` has been incredibly helpful, since I can generally find working python code, and see the bytes, then just translate that into C. I really appreciate this. Keep up the good work!

Start with adafruit_seesaw.rotaryio.IncrementalEncoder and compare its position setter and getter with the captured I2C transactions and the device's rotary register behavior. Done means determining whether the zero readings come from the library, firmware, or byte ordering, and documenting a reproducible position round-trip.

Written by the indexing model from the issue text.