Fix Tensor.tolist()

[yizhuoz004]

memoryview(memref).tolist() does not print the values as expected for some dtypes, such as bfloat16, float8.
Also consider move tolist() and pretty_print APIs to MLIR-TRT's python bindings.

[yizhuoz004]

Also the float16 memoryview cannot return nested lists:

>>> a = np.ones((2, 2), dtype=np.float16)
>>> b = tp.Tensor(a)
>>> b
tensor(
    [1.0000, 1.0000, 1.0000, 1.0000],
    dtype=float16, loc=gpu:0, shape=[2, 2])

[pranavm-nvidia]

@yizhuoz004 does #95 resolve this?

[yizhuoz004]

@yizhuoz004 does #95 resolve this?

No this is still open.

[pranavm-nvidia]

I think #254 addresses this. Do we still want to consider moving tolist and pretty_print to MLIR-TRT?

[yizhuoz004]

Not necessary for now. Let's close this issue.