Add get_from_dict benchmark.

There seems no such existing benchmark.

```
name                                        cpu/op
get_from_dict/size:1/fallback_count:0       1.33µs ± 2%
get_from_dict/size:8/fallback_count:0       1.50µs ± 3%
get_from_dict/size:64/fallback_count:0      2.51µs ± 3%
get_from_dict/size:512/fallback_count:0     10.3µs ± 3%
get_from_dict/size:4096/fallback_count:0    73.5µs ± 3%
get_from_dict/size:10000/fallback_count:0    181µs ± 3%
get_from_dict/size:1/fallback_count:1       2.00µs ± 3%
get_from_dict/size:8/fallback_count:1       2.52µs ± 3%
get_from_dict/size:64/fallback_count:1      5.93µs ± 2%
get_from_dict/size:512/fallback_count:1     33.1µs ± 3%
get_from_dict/size:4096/fallback_count:1     251µs ± 3%
get_from_dict/size:10000/fallback_count:1    625µs ± 3%
get_from_dict/size:1/fallback_count:8       5.40µs ± 2%
get_from_dict/size:8/fallback_count:8       10.7µs ± 2%
get_from_dict/size:64/fallback_count:8      29.2µs ± 1%
get_from_dict/size:512/fallback_count:8      175µs ± 1%
get_from_dict/size:4096/fallback_count:8    1.35ms ± 3%
get_from_dict/size:10000/fallback_count:8   3.29ms ± 2%
get_from_dict/size:1/fallback_count:16      9.23µs ± 2%
get_from_dict/size:8/fallback_count:16      16.6µs ± 2%
get_from_dict/size:64/fallback_count:16     54.8µs ± 2%
get_from_dict/size:512/fallback_count:16     331µs ± 1%
get_from_dict/size:4096/fallback_count:16   2.54ms ± 2%
get_from_dict/size:10000/fallback_count:16  6.13ms ± 1%
```

PiperOrigin-RevId: 721731207
Change-Id: Iaf90b3d7ef549605559b3a205af7b3801cc22e35

Author: goldvitaly

py/koladata/benchmarks.py

@@ -1322,6 +1322,25 @@ def set_get_multiple_attrs_10000_entity_via_fallback(state):
       _ = getattr(merged_ds, arg_name)
 
 
+@google_benchmark.register
+@google_benchmark.option.arg_names(['size', 'fallback_count'])
+@google_benchmark.option.ranges([(1, 10000), (0, 16)])
+def get_from_dict(state):
+  """Benchmark for setting and getting dict key with fallback."""
+  size = state.range(0)
+  fallback_count = state.range(1)
+  ds = kd.dict_shaped(kd.shapes.new([size]))
+  for fb_id in range(fallback_count):
+    ds = ds.with_dict_update(
+        kd.slice([
+            'abc' if i % fallback_count == fb_id else None for i in range(size)
+        ]),
+        kd.slice([fb_id] * size),
+    )
+  while state:
+    _ = ds['abc']
+
+
 @google_benchmark.register
 @google_benchmark.option.arg_names(['nvars'])
 @google_benchmark.option.args([1])