performance: about 5x speedup for inserting into the cache

Seems that `.values()` clause is pretty slow comparing to using `.execute` method instead.

Didn't dig too much why exactly, but the docs kind of give a clue here:

https://docs.sqlalchemy.org/en/14/core/dml.html#sqlalchemy.sql.expression.Insert.values

> It is essential to note that passing multiple values is NOT the same as using traditional executemany() form. The above syntax is a special syntax not typically used.
> To emit an INSERT statement against multiple rows, the normal method is to pass a multiple values list to the Connection.execute() method, which is supported by all database backends and is generally more efficient for a very large number of parameters.

Benchmark:

running `taskset -c 7 pytest -rap --durations=0 -s src/cachew/tests/test_cachew.py -k 'test_many[500000-False]'`

on @karlicoss desktop computer. Using taskset to pin the process to a single core just in case, although that shouldn't matter (note that there is still some jitter, but not sure what it's caused by, might be page faults or something).

baseline -- running without cache at all (@Cachew decorator commented out):

  0.86s
  0.82s
  0.83s
  0.82s
  0.84s

before the change:

  23.29s
  24.44s
  25.25s
  24.41s
  24.46s

after the change:

  5.90s
  5.83s
  5.75s
  5.72s
  5.87s

src/cachew/__init__.py

@@ -985,13 +985,20 @@ def cachew_wrapper(
             # at this point we're guaranteed to have an exclusive write transaction
 
             datas = func(*args, **kwargs)
+            column_names = [c.name for c in table_cache_tmp.columns]
+            insert_into_table_cache_tmp = table_cache_tmp.insert()
 
             chunk: List[Any] = []
             def flush() -> None:
                 nonlocal chunk
                 if len(chunk) > 0:
-                    # pylint: disable=no-value-for-parameter
-                    conn.execute(table_cache_tmp.insert().values(chunk))
+                    # TODO hmm, it really doesn't work unless you zip into a dict first
+                    # maybe should return dicts from binder instead then?
+                    chunk_dict = [
+                        dict(zip(column_names, row))
+                        for row in chunk
+                    ]
+                    conn.execute(insert_into_table_cache_tmp, chunk_dict)
                     chunk = []
 
             for d in datas: