Anthropic: remove limitation "System message must be at beginning of message list."

[nicolablago]

If one tries to call an Anthropic model with multiple system message, the error message reads

"System message must be at beginning of message list."

The check comes from here:

langchain-google/libs/vertexai/langchain_google_vertexai/_anthropic_utils.py

Lines 231 to 232 in 0817f78

	if i != 0:
	raise ValueError("System message must be at beginning of message list.")

Why is this limitation in place?

[mdrxy]

Would be open to a PR

[zamal-db]

I've submitted a fix for this in #1610. The approach removes the positional index check (if i != 0) and instead validates that non-consecutive system messages are not present. Consecutive SystemMessages are merged using the same pattern already used for HumanMessages. This aligns ChatAnthropicVertex with how langchain-anthropic handles system messages. Includes 3 unit tests covering all cases.