feat(python): optimize cython by fastpath

[chaokunyang]

Why?

What does this PR do?

Related issues

Does this PR introduce any user-facing change?

• Does this PR introduce any public API change?
• Does this PR introduce any binary protocol compatibility change?

Benchmark

Before:

================================================================================
SPEEDUP (Fory vs Pickle)
================================================================================
Benchmark                 Fory                 Pickle               Speedup             
--------------------------------------------------------------------------------
dict                      3.99 us              3.81 us              1.05x slower        
large_dict                58.78 us             69.66 us             1.19x               
dict_group                8.38 us              7.78 us              1.08x slower        
tuple                     827.84 ns            587.57 ns            1.41x slower        
large_tuple               3.55 us              2.45 us              1.45x slower        
large_float_tuple         2.97 us              3.35 us              1.13x               
large_boolean_tuple       2.63 us              3.02 us              1.15x               
list                      5.29 us              3.58 us              1.48x slower        
large_list                3.71 us              2.85 us              1.30x slower        
complex                   2.66 us              3.64 us              1.37x  

After:

================================================================================
SPEEDUP (Fory vs Pickle)
================================================================================
Benchmark                 Fory                 Pickle               Speedup             
--------------------------------------------------------------------------------
dict                      3.34 us              3.90 us              1.17x               
large_dict                46.92 us             69.09 us             1.47x               
dict_group                6.63 us              7.77 us              1.17x               
tuple                     597.46 ns            620.41 ns            1.04x               
large_tuple               1.70 us              2.49 us              1.46x               
large_float_tuple         1.85 us              3.37 us              1.82x               
large_boolean_tuple       1.25 us              3.03 us              2.42x               
list                      3.54 us              3.57 us              1.01x               
large_list                1.72 us              2.78 us              1.62x               
complex                   2.50 us              3.68 us              1.47x