docs(SAMPLES-DEC-008): correct the float probe's value count, 47 -> 55

Re-running the probe to restore the source the docs cite showed the count was wrong: it sweeps
55 values (17 named floats, a full Matrix.Identity and a placed transform at 16 each, and 6
doubles), not 47. The result is unchanged -- 55/55 round-trip bit-for-bit -- but a cited number
that does not match what the cited program prints is worse than no number.

Author: robertvokac

docs/XmlSerializationScope.md

@@ -165,9 +165,9 @@ four-space form.
 
 ## Float round-trip: measured, not assumed
 
-`build-probe/xml_probe_float_roundtrip.cpp` swept 47 values — a full `Matrix.Identity`, a placed
+`build-probe/xml_probe_float_roundtrip.cpp` swept 55 values — a full `Matrix.Identity`, a placed
 entity transform, `0.1f`, `1/3`, denormals, `epsilon`, `±INF`, `NaN`, `float`/`double` extremes —
-through `XmlConvert::ToString` → `ToSingle`/`ToDouble` and compared **bit patterns**. 47/47
+through `XmlConvert::ToString` → `ToSingle`/`ToDouble` and compared **bit patterns**. 55/55
 round-tripped exactly. This is why the module adds no float-formatting logic of its own: the
 capability already existed, tested and audited, in `modules/xml`.
 

modules/xml-serialization/include/System/Xml/Serialization/detail/XmlLeafConvert.hpp

@@ -74,7 +74,7 @@ namespace System::Xml::Serialization::detail {
      *
      * Delegates entirely to `System::Xml::XmlConvert`, which is where the float/double
      * round-trip work already lives (probed empirically in `build-probe/
-     * xml_probe_float_roundtrip.cpp`: 47/47 values -- including a full `Matrix.Identity` and a
+     * xml_probe_float_roundtrip.cpp`: 55/55 values -- including a full `Matrix.Identity` and a
      * placed-entity transform -- round-tripped bit-for-bit through `XmlConvert::ToString`/
      * `ToSingle`/`ToDouble`). This header adds no float-formatting logic of its own; it only
      * dispatches to it by type, which is the entire reason Opus's "risk #1" turned out to be