Optimize date parsing in normalize_cansim_values

[mountainMath]

Summary

• Replaces row-by-row as.Date() calls with lookup table approach
• For large tables (millions of rows), date parsing was a major bottleneck
• Builds lookup table from unique date values (typically hundreds/thousands), then uses fast vector indexing

Benchmark Results

Table 14-10-0287 (5,394,600 rows, year-month format):

Version	Mean Time	Min	Max
Original	85.81 sec	82.56	90.02
Optimized	66.97 sec	64.88	69.09
Improvement	22% faster

The optimization saves ~19 seconds on get_cansim() for this large table.

Verification

• Output is identical to original for all tested tables
• All unit tests pass (20/20)

Test Tables

• 20-10-0001 (163K rows, year-month format): IDENTICAL output
• 14-10-0287 (5.4M rows, year-month format): IDENTICAL output
• 34-10-0013 (550 rows, year format): IDENTICAL output

🤖 Generated with Claude Code