Preserve HDR samples and expand high-precision TIFF/EXR support

[JimBobSquarePants]

Prerequisites

• I have written a descriptive pull-request title
• I have verified that there are no overlapping pull-requests open
• I have verified that I am following the existing coding patterns and practice as demonstrated in the repository. These follow strict Stylecop rules 👮.
• I have provided test coverage for my change (where applicable)

Description

Default EXR HALF/FLOAT and floating-point TIFF loads previously passed through bounded pixel storage, losing HDR values. This change selects floating-point pixel storage for those loads and preserves out-of-range samples through the shared pixel conversion path. Associated floating-point pixels also retain stored color at zero alpha.

TIFF encoding now supports IEEE 32-bit floating-point grayscale and color output, plus unsigned 16-bit and 32-bit color output. The TIFF and EXR codec READMEs document their current sample and compression support. SIMD float-plane operations handle bulk conversion, while processor histogram indices remain bounded and percentage thresholds use the observed metric range. This also adds the one-click AutoLevel extension.

Related to #3194.

Validation

• Release solution builds and the pixel-format, codec, SIMD, and processor tests passed on net10.0 and net11.0 before the final README-only commit.
• libtiff decoded 31 ImageSharp-written TIFF files across the supported compression modes; all 9,750 samples matched.
• OpenEXR decoded five ImageSharp-written HDR EXR files; all 1,300 channel samples matched, including nonfinite values and stored color at zero alpha. Its reader also accepted ZIPS output.