AI Activity Improvements and Telegram Bot Stability

Added task comment display in Activity Timeline. Implemented activity display for task commenting in the Activities tab of user profiles. Now showing comment records from Jira and other task boards with the structure "time / label / Commented in (task key + name as link) + comment text," with support for comments from all board types (Jira, Azure DevOps, and others).

Added Meetings to Focus on Activities widget. Implemented meeting display and tracking in the Focus on Activities widget under the Communications section. Activity is created in advance when a meeting is created in Google Calendar with confbot, displays correctly in the Communications label with a percentage relative to other activity types.

Corrected the Meetings weight in the Work vs Communications calculation. Reduced the activity unit weight for meetings by 3x, now giving 1 unit per 15 minutes of call time (0.067 units per minute) instead of the previous inflated value. This eliminated the critical discrepancy between the actual activity list in Activity logs (where Tasks/Work predominated) and the Focus on Activities widget (which showed Communications dominance at ~72%), and ensured correct weight ratios across different activity types.

Fixed activity unit calculation in the Focus on Tasks widget. Eliminated the calculation error in Bug vs. Feature distribution in the Focus on Tasks widget; fixed the percentage ratio formula based on worklogs and activity weights. Now with 3 tasks (Features) with worklogs and 1 bug with worklogs, the ratio correctly displays as Bugs 25% / Features 75% instead of incorrect values, ensuring proper operation for any number of tasks of different types. You can now correctly identify technical debt situations (when >50% of time goes to bugs) or insufficient quality attention (when <10% of time on bugs with growing backlog), which is critical for team prioritization and resource decisions.

Removed excessive text thank-you messages from the Telegram bot for standups. Removed the annoying text thank-you function from the Telegram bot; now the bot only adds a reaction (like) when receiving a correct standup instead of sending a text message like "Thanks!" or similar. Text feedback will only be sent when the standup format is incorrect to inform the user of the error.

Fixed critical Telegram bot errors with missing usernames and special characters. Eliminated three Telegram standup bot problems:

1. Fixed handling of users without a Telegram username: the bot correctly accepts standups and doesn't add "@None" to messages.
2. Accelerated Chat/Topic ID data transmission to Enji: now happens immediately when bot is tagged in a new chat/subgroup instead of being delayed.
3. Fixed special character parsing in topic names: correctly handles brackets and ampersands (<, >, &) when renaming topics.

Impact: Our clients now have a stable bot operation in real-world conditions. Eliminated the blocking issue for users without usernames (old accounts registered only with phone numbers) who couldn't submit standups and received incorrect "@None" mentions in alerts; accelerated new chat/topic registration (instant ID transmission instead of delay); ensured correct operation with real topic names containing special characters (e.g., "Dev & API"). This is critical for enterprise clients with diverse user bases and structured Telegram chats.

Fixed employee feedback formatting and removed automatic capitalization. Eliminated excessive automatic capitalization when writing employee feedback; each new word no longer starts with a capital letter automatically. Now, when filling in data, the user controls text case, and text field behavior has been fixed for natural input.