Consumer docs disagree with the 10.0.0 code in five places #278

Problem

A docs-against-code check on main e465c45 compiled and ran every sample about a 10.0.0 addition. All of them matched except the items below.

1. docs/serialization.md:194: the span sample throws. line[7..29] of "period=[2024-01-01,2024-12-31];rate=4.5" cuts off the closing ], so DateRange.Parse throws FormatException. It should be line[7..30].
2. src/CodoMetis.ValueRanges.EFCore.PostgreSQL/README.md, "What runs in memory only": the README says "Everything else … runs in the database". That's false for Shift, Expand, Split, Values(step) and the calendar factories over a column. docs/efcore.md "What runs where" and the changelog both list those as in memory only. The README's operator table also leaves out CompareTo → </<=/… and the range-with-multirange operand.
3. src/CodoMetis.ValueRanges/README.md:9: says "the six element types … plus TimeRange", but its own table (lines 63–64) lists Float64Range and Float32Range. The NuGet <Description> in the core and EF csproj files leaves out the float family and the 10.0.0 additions, and that text shows on nuget.org.
4. src/CodoMetis.ValueRanges.NodaTime/README.md, "Use them with EF Core": says the RangeAgg/RangeIntersectAgg overloads belong to "the EF Core NodaTime package". They're declared in the NodaTime package (NodaTimeRangeAggregateExtensions.cs); the EF satellite only registers them.
5. docs/architecture.md:449 (agent doc): names DoubleRange as the worked example of Float64Range/Float64Set and Float32Range/Float32Set over finite values only, mapped to custom float8range and float4range #153. The type is Float64Range, and the sibling-axis table on line 459 leaves the float ranges out.

Optional: docs/ranges.md:707 and docs/efcore.md:642 say "PostgreSQL 14+" about multiranges. That's historically true, but next to the 16 floor (#215) it can read as if 14 were supported.

Done when

All five are fixed. Item 1 is covered by a claim or a test, so the sample can't drift again.