break(helpers): use the exact defined conversion factors for feet, inches, yards, acres and square miles

[dylanpulver]

factors and areaFactors in @turf/helpers already use exact defined values for metres, kilometres, miles (earthRadius / 1609.344) and nautical miles (earthRadius / 1852). The imperial entries
are rounded literals instead, and they disagree with each other.

The 1959 international yard and pound agreement fixes 1 yd = 0.9144 m, 1 ft = 0.3048 m and 1 in = 0.0254 m exactly, so 1 mile = 1760 yd = 5280 ft holds by definition. On master it does not:

convertLength(1, "miles", "yards")      1759.9785983999998    should be 1760
convertLength(1, "yards", "feet")       3.0000365764447703    should be 3
convertLength(1, "feet",  "inches")     11.999975616000778    should be 12
convertArea(1,   "miles", "kilometers") 2.5906735751295336    should be 2.589988110336
convertArea(1,   "miles", "acres")      640.1683937823834     should be 640

inches: 39.37 is not a rounding: it is the exact US survey inch (1 US survey ft = 1200/3937 m), which NIST and NGS retired for federal use after 2022-12-31 — and 39.37 / 12 = 3.280833… does not match the neighbouring feet: 3.28084. The largest error is areaFactors.miles = 3.86e-7, 2.65e-4 relative: a 100 km² polygon reports 38.6 sq mi instead of 38.610215854…

The fix writes every by-definition-exact factor as a division by its defining SI value, in the style the table already uses for miles and nautical miles. areaFactors.nauticalmiles was already bit-identical to 1 / 3429904 and is left alone. Precedent: #2177 corrected the degrees factor in this same table. No open issue tracks this; #1857 and #1924 fixed the inverted yards factor for v6.4.0 but kept the rounded literal.

Tests. Eight existing assertions pinned the old constants — convertLength(1, "yards", "meters") asserted 0.91441111923921, where the defining constant is 0.9144 — and are updated. Two new tests assert the definitional identities (1 yd = 3 ft = 36 in, 1 mi = 1760 yd, 1 acre = 4840 sq yd, 1 sq mi = 640 acres) rather than factor values, so they cannot be satisfied by whatever the tables happen to hold. They allow 4 ulp; the area identities land bit-exact.

Verification. With index.ts reverted to master, 9 of the 11 new assertions fail; the 2 that pass are nautical miles and hectares, which were already exact, and are there as a control. Correcting only factors leaves the four imperial area identities failing; correcting only areaFactors leaves the five imperial length identities failing. Whole-repo suite is identical before and after: 112 packages pass, and turf-directional-mean and turf-standard-deviational-ellipse fail on unmodified master here too (Node 24, float drift in the 11th digit of bearingAngle) — please compare against your own baseline.

Breaking: yes. Distances in feet, inches or yards, and areas in square miles or acres, change in the 5th to 8th significant figure.

Please provide the following when creating a PR:

• Meaningful title, including the name of the package being modified.
• Summary of the changes.
• Heads up if this is a breaking change.
• Any issues this resolves.
• Inclusion of your details in the contributors field of package.json - you've earned it! 👏
• Confirmation that you've read the steps for preparing a pull request.