Fix Fraction.add and subtract for operands not in lowest terms

[lenamonj]

Fraction.addSub runs Knuth 4.5.1 on the operands as given, but the algorithm assumes lowest terms and getFraction(50, 100) is stored as 50/100. A shared factor inside an operand survives into the common denominator, so add and subtract throw ArithmeticException while the exact result fits an int (1073741823/2147483646 plus 3/5 is 11/10) and return unreduced results against the reduced form the javadoc promises. #1769 fixed the same shape in multiplyBy.

Both operands are now reduced before the Knuth step, as multiplyBy does. Operands already in lowest terms take the identical path, so the existing overflow and identity assertions hold.

Test: testAddSubtractUnreducedOperands, red on master and green with the change. Default Maven goal green.

• Read the contribution guidelines for this project.
• Read the ASF Generative Tooling Guidance if you use Artificial Intelligence (AI).
• I used AI to create any part of, or all of, this pull request. Which AI tool was used to create this pull request, and to what extent did it contribute? Claude (Anthropic) found the defect and drafted the fix and test; I reviewed them, reproduced the failure on a fresh clone of master, and ran the default Maven goal.
• Run a successful build using the default Maven goal with mvn; that's mvn on the command line by itself.
• Write unit tests that match behavioral changes, where the tests fail if the changes to the runtime are not applied. This may not always be possible, but it is a best practice.
• Write a pull request description that is detailed enough to understand what the pull request does, how, and why.
• Each commit in the pull request should have a meaningful subject line and body. Note that a maintainer may squash commits during the merge process.

[garydgregory]

Hi @lenamonj
Thank you for the PR, the main changes look good but I think you need more tests, specifically:

• Extend the regression tests for subtraction. The new test covers both denominator branches for addition, but its only subtraction example has coprime denominators after reduction. Add an unreduced subtraction case with shared reduced denominators, for example 2/4 - 2/12 = 1/3, and a cancellation case such as 2/4 - 3/6 = 0/1.
• Exercise both operands being unreduced and minimum-integer normalization. Examples such as 2/4 + 2/6 = 5/6 and Integer.MIN_VALUE/2 + 2/4 = -2147483647/2 directly protect the two new normalization steps and their signed boundary behavior.
• Bonus: Clarify the existing reduced-result promise around zero. The new normalization runs after the zero shortcuts in addSub (line 550). Consequently, ZERO.add(2/4) still returns 2/4. Also, ZERO.subtract(Integer.MIN_VALUE/2) still throws even though its reduced result, 1073741824/1, fits. These are pre-existing limitations, not regressions.

Thank you!

[garydgregory]

Thank you @lenamonj , looks good, merged 🚀