W-23900223 : Hyper Query Plan - Replace raw property dumps with tidy node rows.

[rbrahmachary]

Hyper Query Plan - Replace raw property dumps with tidy node rows.
Node bodies previously dumped debug-name, statistics, and sqlpos verbatim via forceToString, producing noisy JSON blobs. Surface the useful figures as clean, curated rows instead.

• Replace the raw dump with curated rows: table-or-alias (from debug-name.value), estimated-rows, output-rows, processed-rows, rows-matching, memory-bytes, cpu-cycles, execution-time
• New : Add a grouped table-metadata block (identifier / partitioned-by /sort-order) for lakehouse scans, rendered as a header plus sub-items
• Show memory-bytes in binary units via a new formatBytes helper (KiB/MiB/GiB), clearer than the decimal formatMetric
• Display row order: identity/volume first, operator identifiers last
• Echo the runtime hotspot flag onto the cpu-cycles row, reusing the node heat color so the tint intensity matches the label
• Add getStatistic / getEstimatedRows / getActualRows helpers that read statistics and fall back to legacy analyze / cardinality
• Add a firstNumber helper that skips non-numeric values instead of masking a real number further down the fallback chain
• Reuse those helpers for the edge label, hotspot heatmap, and errored-icon detection, keeping tooltip figures and visual cues consistent (including on legacy-format plans)
• Extract the hotspot and cardinality-mismatch magic numbers into constants
• Add groupedProperties / highlightedProperties to TreeNode and the grouped-header / highlighted-row rendering in QueryNode

Query Graph node rows:-