Interpret scalar values using their declared types

The same bits can have different meanings; scalar type constraints still apply.

• A signed int is 32-bit two’s complement in−2147483648…2147483647. An unsigned int is 32 bits in 0…4294967295. Bytes ff ff ff fe therefore mean −2 under int and 4294967294 under unsigned int; the schema chooses the interpretation (§§4.1–4.2).
• An enum has the signed-int representation, but encoding an unassigned value is an error. Boolean is the enum FALSE=0,TRUE=1, still four bytes; arbitrary nonzero integers are not true. In the RFC’s illustrative color enum, only 2,3,5 are assigned, so4 is not legal (§§4.3–4.4). These are separate type examples, not one combined type.
• Hyper and unsigned hyper are64-bit, eight-byte integers, most significant byte first. Signed hyper extends two’s complement; unsigned hyper covers0…2^64−1. The example4294967296 uses00 00 00 01 00 00 00 00. The teaching helper uses BigInt to avoid rounding large integers through JavaScript Number (§4.5).
• Float is 32 bits: sign1 bit, exponent8 bits biased by 127, fraction23 bits. The normalized formula is(−1)^S ×2^(E−bias) ×1.F. The worked value 3.5 has S=0,E=128,F=0x600000 and bytes40 60 00 00 (§4.6). Do not apply that normalized formula blindly to zero, infinity, subnormal or NaN encodings.
• Double is 64 bits with fields1/11/52 and bias1023. Quadruple is 128 bits with fields1/15/112 and bias16383. The final state compares actual bit spans, not host arithmetic support. Sections4.6–4.8 and 11 address special values; a NaN is interpreted only as NaN, not a portable payload or application value. The helper does not implement quadruple arithmetic.