Column design: axial compression and combined axial plus flexure (SNI 1729 ch. E and H) #1

The crate checks beams. It cannot check a column, which is the other half of any frame and the part that fails brittly.

What is missing

• Axial compression, SNI 1729 ch. E: flexural buckling, torsional and flexural-torsional buckling, the effective length KL/r, and the Fcr branches either side of the slenderness limit.
• Combined axial and flexure, ch. H: the Pr/Pc interaction, which is where a column actually gets sized.
• Local buckling of the compression elements, since the existing classification is written for flexure.

Why it matters more than it looks

A beam that is 5% overstressed sags and someone notices. A column that is 5% overstressed fails without warning. This is the module where returning an error instead of a number matters most, so the existing habit of refusing unimplemented cases should carry over: no approximate Fcr for a section shape that is not covered.

Suggested order

1. check_compression for a doubly symmetric rolled I section, flexural buckling only, with K as a caller input rather than guessed from a frame the library cannot see.
2. Torsional and flexural-torsional buckling.
3. The ch. H interaction, once compression exists.

Step 1 alone is useful and is roughly the size of the existing flexure module. Worked-example tests with the hand calculation in a comment, same as the rest of the crate.