Nnio

Changed [nnio.DetectionBox](https://nnio.readthedocs.io/en/latest/utils.html#nnio-detectionbox) class:
* Attributes are now named `x_min, y_min, x_max, y_max`
* `x_min` and `x_max` are for width axis and `y_min`, `y_max` are for height axis, like in PASCAL VOC.

Changed file caching. Now files from different urls will not conflict when they have the same name.

New models for person re-identification:

* `nnio.zoo.onnx.reid.OSNet`
* `nnio.zoo.openvino.reid.OSNet`
* `nnio.zoo.edgetpu.reid.OSNet`

New class: `nnio.utils.HumanDataBase` for using with person re-id models.

nnio.zoo.edgetpu.DeepLabV3` model is now the light version.

* Now python's auto completion will not show non-relevant imports, especially for `nnio.zoo`
* Cache is now saved in `/home/user/.cache/nnio/0.2/`. Earlier it was saved to `.../nnio/0.2.1/`. Now small updates will not affect saved models.
* Added `__eq__` operator for `nnio.Preprocessing`.

More convenient device naming
* Names for all devices are written in the same form:
`CPU`, `GPU`, `MYRIAD`, `TPU`
* When multiple myriads or TPU's are available, user can choose between them as:
`MYRIAD:0`, `MYRIAD:1`, `TPU:0`, `TPU:1`, ...

Measuring myriad temperature
* Temperature will be returned from the model call if `return_info` flag is used:
`output, info = model(image_prepared, return_info=True)`
* Temperature of each myriad can be logged to `/home/$USER/.telemetry` after each model call.
To enable this:
`nnio.utils.enable_logging_temperature(True)`

New ready models:
- nnio.zoo.edgetpu.segmentation.DeepLabV3
- nnio.zoo.onnx.classification.MobileNetV2
- nnio.zoo.openvino.detection.SSDMobileNetV2