`tool_call_id` truncation error during tool calling with Gemini models

**Title:** `tool_call_id` truncation error during tool calling with Gemini models

**Description:**
When using Google Gemini models (e.g., `gemini-2.5-flash`) for tool calling, the session crashes with a truncation error. Gemini generates significantly longer `tool_call_id` strings than OpenAI, which exceed Frappe's default 140-character limit for `Data` fields.

Because the database insert fails, the flow's state is marked as `Failed`, preventing it from being resumed.

**Steps to Reproduce:**
1. Configure a Google AI Studio / Gemini model in the Flow Model DocType.
2. Create a Flow Session and trigger a prompt that requires a tool/function call.
3. The session crashes when attempting to save the `Flow Session Message`.

**Error Log:**
`Error: Flow Session Message, Row 7: 'Tool Call ID' (call_a9efe2adc63948f48e20943b87cethoughtEl4KXAERTTIPeUBLJ5fdyJDdi++R1Hr1st/9pRsw+sCVQRSKxATqH4CnIHJaEqw0Vzo1EwwTQ6PYT3/VVfXXi2W+9a+gWaNFno/1WVc/x/ulbv1WYlZGMCIA8kJ0CwqG) will get truncated, as max characters allowed is 140`
`Error: Only Paused runs can be resumed (this run is Failed).`

**Expected Behavior:**
The system should accommodate longer Tool Call IDs from non-OpenAI providers to allow successful execution and state saving.

**Suggested Fix:**
In the `Flow Session Message` DocType, change the field type of `tool_call_id` from `Data` to `Small Text`, or increase the `Data` length property to 255+ to safely handle Gemini's verbose ID generation.

Start with the Flow Session Message DocType and inspect the tool_call_id field, currently described as a Data field with a 140-character limit. Reproduce a tool call using a Google Gemini model such as gemini-2.5-flash, then verify that the Flow Session Message saves successfully and the session is not marked Failed or prevented from resuming.

Written by the indexing model from the issue text.