feat: add worklog analytics tools and harden auth/error handling

[ivelin-web]

Summary

Adds two new MCP tools for worklog insights, plus several bug fixes that surfaced while wiring them up.

New tools

• getMissingWorklogDays — finds working days where logged time is below the user's expected schedule. Uses Tempo's user-schedule API so holidays, non-working days, and part-time schedules are honoured automatically. Per-issue breakdown for partially-logged days.
• getWorklogAnalytics — aggregates worklogs by issue / account / day / week (ISO 8601) / month. Returns hours, percentage, worklog count per group sorted by hours desc.

Bug fixes (surfaced during integration)

• getCurrentUserAccountId now prefers /rest/api/3/myself for all auth types, with the email-search retained as a basic-auth fallback. Avoids empty results when Atlassian privacy or scoped-token restrictions hide email visibility.
• All 7 tool wrappers now propagate isError correctly. bulkCreateWorklogs was especially affected — full-batch failures were silently reported as successes because the wrapper was dropping its explicit isError flag.
• Pagination MAX_PAGES limit now throws instead of silent truncation. Returning incomplete data was worse than failing loudly.
• 403 from /user-schedule surfaces a clear scope-guidance message instead of an opaque axios error.

Polish & optimisation

• limit=1000 on Tempo paginated requests — typical month-long queries now fit in a single round trip.
• formatHours / formatPercent helpers drop trailing zeros (7.5h, 50%).
• validateDateRange rejects startDate > endDate with a clear MCP error.
• Cleaner output: single text block per response, no leading dashes (which clashed visually with em-dashes in issue titles), two-line layout for issue grouping.
• Consolidated getIssueInfoMap helper (resolves key + summary in one pass); removed dead getIssueKeysMap / getIssueKeyById.

Docs

• README clarifies classic vs scoped Jira API tokens and the OAuth fallback for orgs that disable classic tokens.
• .env.example notes the Schemes scope requirement for Tempo.
• Tool descriptions added in index.ts including the account groupBy caveat.

Bumps to 1.6.0.

Test plan

• npm run build clean
• npm run lint — 0 errors
• npm run format:check clean
• getMissingWorklogDays tested against live Tempo: zero-logged days, partially-logged days with breakdown, full week with no missing
• getWorklogAnalytics tested with groupBy: issue (returns key + Jira summary, percentages sum to 100%)
• 403 scope error verified — gives clear guidance message
• Empty schedule case — returns informative message, not error
• OAuth and bearer auth paths not tested in this PR (changed via /myself preference; existing logic preserved)
• Inverted date range (startDate > endDate) verified to short-circuit with the new guard
• bulkCreateWorklogs full-batch failure — verify isError now reaches the MCP client (existing behaviour was silently success)

Notes for reviewers

• The "Create API token with scopes" Atlassian flow does not work with this server's basic auth path (scoped tokens require the api.atlassian.com/ex/jira/{cloudId} gateway). README now calls this out and points to OAuth as the workaround. A follow-up PR could add native scoped-token support via a JIRA_CLOUD_ID env var.
• The account groupBy in getWorklogAnalytics only makes sense if your team uses Tempo accounts (_Account_ work attribute on worklogs). Documented in the tool description.