Changes in version 0.1.0.9000 - Retry a fresh descendant snapshot when a child vanishes during Linux process enumeration. Preserve observed process identities and propagate repeated or unrelated inspection errors; tests retain child identity before it exits. - Track and verify observed worker descendants during timeout, failure and successful completion; isolate output streams from inherited pipes and bound receipt logs. Native slide reads use the same process handling. - Preserve comparison run names and excluded physical-slide IDs in canonical deliveries and display comparison direction in HTML reports. - Retain valid results when a saved refresh loses its configured grouping column, and reject annotation transforms that overflow numeric range. - Standardize the named v5 package logo across README, documentation pages and available application headers, with matching light and dark artwork. - Preserve exact tissue boundaries across platforms and distinguish exported table names during folder imports. - Re-audit: exact count parsing and identity checks in Shiny, unambiguous filters, preserved comparison exports, and catchable bloodspottr_error conditions. - Added preview/native whole-slide browsing, QuPath GeoJSON and point-CSV selection, saved-file refresh, arbitrary tissue labels and configurable grouping. - Added a Shiny tutorial, responsive branded layout, completeness explanation, and shared CTTIR check, coverage and blocking house-lint workflows. - Add validated legacy and canonical slide-result import, pooled burden summaries, and comparisons that require matching measured support. - Add immutable delivery directories with CSV, JSON, optional XLSX, HTML reports, and SHA-256 manifests. - Add a Shiny results explorer with explicit synthetic demonstration data, organ and quality filters, missing-data visibility, and guarded uploads. - Add a bounded review queue with saved-export receipts and class-specific scope. - Add a versioned external-worker protocol with timeout cleanup and input/output verification; native DNN workers remain separately qualified integrations. - Add source-suite inventory and calibrated candidate exchange with cellspecR. - Preserve the existing calibrated burden and spatial APIs. - Add relative OD/chromaticity features and calibrated directional/profile diameter flags, with no area-based cell-count inference.