[Bug]: HTTP SDK corrupts varchar values that look like booleans ("true"/"false"/"None") in to_df

### Is there an existing issue for the same bug?

- [x] I have checked the existing issues.

### Version or Commit ID

main @ 4954148 (current main as of 2026-09-08)

### Other environment information

```Markdown

```

### Actual behavior and How to reproduce it

The HTTP server serializes boolean columns as the strings "true"/"false" (default branch of the value switch in HTTPSearch::Search -> ColumnVector::ToString(row)), so the HTTP Python client (table_http_result.to_result in python/infinity_sdk/infinity/infinity_http.py) coerces matching strings back to booleans. But it does this for ANY string value without checking the column's declared type:

if v.lower() == 'true':
v = True
elif v.lower() == 'false':
v = False
elif v.lower() == 'none':
v = None

So a varchar cell containing "true" comes back as True (rendered "True" in the string DataFrame column), "false" as False, and "None" as a null - stored data is corrupted on read. Nulls are already carried by the _bitmap fields, so this coercion is only ever needed for actual boolean columns.

Repro (stub the net layer, feed the exact JSON shape the server produces):

SERVER_JSON = {"error_code": 0, "output": [
[{"name": "true", "name_bitmap": True}, {"flag": "true", "flag_bitmap": True}],
[{"name": "None", "name_bitmap": True}, {"flag": "false", "flag_bitmap": True}],
]}
# table columns: name varchar, flag bool
# to_df() -> name: ["True", ] (expected: ["true", "None"])
# flag: [True, False] (correct)

### Expected behavior

Varchar data passes through verbatim: "true" stays "true", "None" stays "None". Only columns whose declared type is boolean (and untyped expression results, e.g. "c1 > c2") get the "true"/"false" string coercion.

### Additional information

Fix: gate the coercion on the declared column type from show_columns_type() (keeping it for expression columns with no declared type). PR with the fix plus a regression test (round-trips "true"/"None"/"false" through a varchar column) coming.

Start in python/infinity_sdk/infinity/infinity_http.py at table_http_result.to_result, then inspect how show_columns_type() supplies declared types. Reproduce the issue with the provided SERVER_JSON shape and add a regression test covering varchar values "true", "None", and "false"; done means those strings pass through while boolean columns and untyped expression results retain their expected coercion.

Written by the indexing model from the issue text.