Adapt Conversations Adapter

You are given a JavaScript function named `createOngoingConversationsObject` that takes an array of conversation objects and a locale parameter, and returns a transformed array of conversation summaries. The function processes an input array of conversations that may contain varying object structures, including a "new conversation shape" with some different property arrangements. Your task is to modify this adapter function so that it correctly supports and adapts both the existing conversation objects and the new conversation shape, merging them seamlessly into a unified structure with the following fields for each conversation: - `id`: conversation ID - `createdAt`: conversation creation timestamp - `isPartnerChat`: boolean, true if the conversation is from an agent or partner (false if from client) - `user`: an object containing - `avatar`: user's avatar URL or empty string - `fullName`: user's name or empty string - `isOnline`: user's online status boolean - `id`: user ID - `lastPage`: last page URL - `clientId`: ID of the client user in the conversation - `notSeenMessages`: number of unseen messages for the widget - `lastMessage`: last chat message text - `messageTime`: last message update time (usually string like '۱۷:۵۲' or similar) - `updatedAtSort`: timestamp string parsable to a Date object - `channelId`: the channel ID of the conversation Requirements: - The adapted function should handle conversations array elements that may be missing some fields or have different nested structures in the "new shape". - For the new conversation shape, ensure you extract and map relevant fields correctly from their different locations. - Retain fallback defaults (empty strings, zeroes, booleans) when any field is missing. - The `updatedAtSort` field should parse time correctly via the existing utility `convertTimeToDateTime` where possible. - Preserve correct typing: booleans, strings, IDs, numbers. - Avoid mutation of input objects; build a new adapted array. - The function should be robust, readable, and maintainable. # Steps 1. Analyze both conversation shapes (existing and new). 2. Establish rules to detect each shape and extract required fields accordingly. 3. Refactor the adapter loop to handle either shape. 4. Normalize user info and conversation metadata to match output schema. 5. Return the unified adapted array. # Output Format Return the full modified JavaScript function code for `createOngoingConversationsObject` accepting `(conversations = [], locale)` as parameters and returning the adapted array as described. Include all necessary internal helper uses such as `convertTimeToDateTime`. Do not alter unrelated logic.