Fix stuck export/upload progress caused by unindexed logs table

[caitmich]

Summary

While diagnosing a hosted Pro instance with 47.2 million rows in its logs table, we found the root cause is present in CE too: the logs table has no index beyond its primary key, and the client-side progress-polling JS has no error handling. Together these mean that once a logs table grows large enough for lookups to slow down, progress polling can silently and permanently stop updating, with no indication to the user that anything went wrong.

• db/migrate/..._add_index_to_logs_on_uid.rb: adds an index on logs.uid, the column ConsoleController#status filters on for every progress-polling request. Without it, each poll is a full table scan. Note: the equivalent Pro migration uses algorithm: :inplace to force an online, non-locking index build on MySQL, but that option isn't supported by the SQLite3 adapter, which CE defaults to — passing it here breaks migrations outright for most CE installs, so it's omitted for this version.
• app/assets/javascripts/hera/modules/console_updater.js.coffee: the polling loop only rescheduled its next check from inside the AJAX success callback, with no .fail() handler — a single dropped/timed-out request permanently stopped progress updates with no error shown. Now retries on failure (resetting the retry count on any success, so intermittent blips don't accumulate), and after 5 consecutive failures stops retrying and shows a "lost connection, please refresh" message instead of hanging silently forever.
• CHANGELOG: entries for the above.

This will be synced into Pro (which has its own equivalent fixes plus two Pro-only changes: batching LogCleanupJob's deletion, and the same polling fix in the Word-export validator).

Testing steps

Confirm the logs table index exists after migrating
Trigger an export or upload and confirm progress updates appear live in the console without needing a manual refresh
Simulate a failed progress-polling request (e.g. block the endpoint temporarily) and confirm polling retries automatically, then shows the "lost connection" message after repeated failures instead of hanging indefinitely

Check List

• Added a CHANGELOG entry
• Commit message has a detailed description of what changed and why.