MobileBERT is ExecuTorch compatible

[guangy10]

What does this PR do?

MobileBERT is ExecuTorch compatible.

Unit Test:
RUN_SLOW=1 pytest tests/models/bert/test_modeling_mobilebert.py -k test_export -v

tests/models/mobilebert/test_modeling_mobilebert.py::MobileBertModelIntegrationTests::test_export PASSED                                                                                [100%]

E2E test in ExecuTorch:
Patch pytorch/executorch#6509
python -m extension.export_util.export_hf_model -hfm="google/mobilebert-uncased" -lm="masked_lm"

Saved exported program to ./mobilebert.pte

./cmake-out/backends/xnnpack/xnn_executor_runner --model_path mobilebert.pte

I 00:00:00.052851 executorch:executor_runner.cpp:82] Model file mobilebert.pte is loaded.
I 00:00:00.052931 executorch:executor_runner.cpp:91] Using method forward
I 00:00:00.052941 executorch:executor_runner.cpp:138] Setting up planned buffer 0, size 78562304.
I 00:00:00.080190 executorch:executor_runner.cpp:161] Method loaded.
I 00:00:00.080223 executorch:executor_runner.cpp:171] Inputs prepared.
I 00:00:01.315332 executorch:executor_runner.cpp:180] Model executed successfully.
I 00:00:01.315349 executorch:executor_runner.cpp:184] 1 outputs:
Output 0: tensor(sizes=[1, 64, 30522], [
  -8.54569, -9.2325, -9.63534, -9.15334, -8.9625, -8.23703, -8.13457, -8.46437, -9.27599, -10.4376,
  -9.40925, -8.32582, -7.68392, -9.18805, -9.3336, -8.71841, -9.52881, -8.03673, -10.02, -9.08982,
  -10.1946, -9.01746, -8.93738, -8.80435, -9.85641, -10.1131, -8.68367, -7.66197, -11.1961, -9.6375,
  -10.6121, -9.54771, -9.9602, -9.8043, -8.37304, -9.33157, -10.0574, -9.67762, -8.67488, -9.94721,
  -8.77431, -11.054, -9.39442, -9.91193, -8.54864, -10.2091, -9.69447, -10.2535, -8.71598, -8.85885,
  -10.5134, -10.151, -9.6825, -10.2545, -9.03317, -9.3754, -10.1931, -8.01357, -7.8706, -9.44086,
  -7.66918, -8.66947, -8.13562, -8.58915, -11.6404, -8.62273, -8.92262, -9.81593, -8.25118, -8.82181,
  -11.4176, -9.58474, -8.92113, -8.37405, -8.7471, -8.637, -9.41958, -8.99523, -10.1699, -9.5519,
  -8.48485, -9.90429, -8.82682, -9.10106, -9.65399, -9.55595, -9.16413, -7.88924, -8.53024, -8.77796,
  -9.7509, -9.06045, -10.9217, -10.7008, -8.35735, -10.2375, -9.26975, -8.89627, -9.15771, -9.24441,
  ...,
  -7.41324, -7.98777, -6.96755, -5.78335, -6.45162, -6.61997, -7.14526, -7.84212, -7.80403, -6.78576,
  -6.23739, -6.91561, -5.49068, -6.96955, -6.45688, -6.2613, -5.91608, -6.87685, -6.08245, -6.84436,
  -7.63106, -6.86492, -7.27272, -7.66639, -5.61715, -6.88419, -6.54495, -7.39954, -4.94468, -7.13938,
  -6.0773, -6.17809, -6.97764, -7.4307, -6.80996, -6.7768, -7.37503, -7.60815, -7.17836, -6.44587,
  -7.20048, -5.83285, -6.8438, -5.48847, -6.91492, -6.98491, -5.9548, -7.00031, -6.17852, -6.701,
  -6.21065, -7.01716, -7.04772, -7.45687, -7.57249, -5.39035, -6.14504, -6.50262, -6.42711, -7.29622,
  -6.24016, -7.12543, -7.03201, -7.88257, -7.04191, -7.0761, -6.95057, -6.0472, -5.97595, -6.85439,
  -5.8476, -6.37267, -6.20555, -7.30387, -6.25161, -7.53724, -6.54707, -6.81088, -6.88549, -6.02146,
  -6.34617, -6.92539, -7.22334, -7.04436, -6.80034, -6.05272, -6.70695, -8.43641, -0.241071, -2.15037,
  -6.41449, -5.42762, -6.04412, -7.70756, -7.53018, -6.31514, -7.02132, -6.56385, -7.70365, 0.30994,
])

Before submitting

• Did you read the contributor guideline,
  Pull Request section?
• Was this discussed/approved via a Github issue or the forum? Please add a link
  to it if that's the case. MobileBERT is ExecuTorch compatible #33843
• Did you write any new necessary tests?

Who can review?

@ArthurZucker
@qubvel

[ArthurZucker]

LGTM thanks