Fix issue with nested model output as input

[Surya2k1]

Currently with nested model output as input to functional model fails as tensor_dict not updated. The below code fails.

model_1 = Sequential([
    Input(shape=(6,)),
    layers.Dense(3, activation="sigmoid"),
])

model_2 = Sequential([
    Input(shape=(3,)),
    layers.Dense(1, activation="sigmoid"),
], )

combined = Model(Input(shape=(6,)), model_2(model_1(Input(shape=(6,)))), name='nested_model')
combined.compile(loss='binary_crossentropy', optimizer='adam')
output = combined.train_on_batch(np.random.normal(0, 1, (8, 6)), np.random.normal(0, 1, (8, 1)))

Done the changes to make it work not seems elegant though. But for this case it works and no functional model tests are failing with this change.

Context from #20473

[codecov-commenter]

Codecov Report

Attention: Patch coverage is 42.85714% with 4 lines in your changes missing coverage. Please review.

Project coverage is 72.12%. Comparing base (3dd958b) to head (55e741e).

Files with missing lines	Patch %	Lines
keras/src/ops/function.py	42.85%	2 Missing and 2 partials ⚠️

❗ There is a different number of reports uploaded between BASE (3dd958b) and HEAD (55e741e). Click for more details.

HEAD has 4 uploads less than BASE

Flag	BASE (3dd958b)	HEAD (55e741e)
keras	5	3
keras-numpy	1	0
keras-torch	1	0

Additional details and impacted files

@@            Coverage Diff             @@
##           master   #20678      +/-   ##
==========================================
- Coverage   81.97%   72.12%   -9.86%     
==========================================
  Files         544      544              
  Lines       50751    50756       +5     
  Branches     7851     7854       +3     
==========================================
- Hits        41604    36607    -4997     
- Misses       7252    12364    +5112     
+ Partials     1895     1785     -110     

Flag	Coverage Δ
keras	72.05% <42.85%> (-9.77%)	⬇️
keras-jax	63.95% <42.85%> (+<0.01%)	⬆️
keras-numpy	?
keras-openvino	29.93% <14.28%> (-0.01%)	⬇️
keras-tensorflow	64.70% <42.85%> (+<0.01%)	⬆️
keras-torch	?

Flags with carried forward coverage won't be shown. Click here to find out more.

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.