docs: sync DM hardware BOM sections across all five languages

[Chirag6722]

Refs #64 (items 1–3; items under 4 left open deliberately — see below)

What this changes

hardware/reBot_B601_DM/ is the document people build from. Its five language versions had diverged enough that what you can build depended on which one you read:

	readme.md	readme_es.md	readme_zh.md	readme_fr.md	readme_jp.md
tables — before	11	11	12	8	8
tables — after	12	12	12	12	12
links — before	31	31	35	9	29
links — after	32	32	35	32	32

1. French and Japanese had no power-supply assembly section

Both stopped after a superseded 3-row BOM — still listing a 12AWG cord and an XT30 16AWG cable, where the current BOM specifies an LRS-350-24, an XT60E output, a 3-in-1 IEC inlet socket and the XT30U-to-XT60 adapter. This PR ports the consumables BOM (10 rows), the printed-parts BOM (3 rows) and the 12-step illustrated procedure with its 28 step images.

This is mains-voltage wiring, and readers of those two languages had a superseded parts list with no assembly guidance for it. The content arrived in #33, #37, #48 and was corrected in #54 — each time in English only.

2. The French BOM had lost its purchase links

20 rows of the purchased-parts table read Amazon as plain text. Relinked from the corresponding English rows.

3. The camera-mount section was missing from English and Spanish

It existed in French, Japanese and Chinese. Ported to both, so the drift is corrected in the direction it actually runs rather than only away from English.

4. All three existing copies named a STEP file that isn't in the repository

They listed Gemini2_mount.step; the file is D435_Gemini2_Mount.step. All five now point at the real file, as a link, so the link checker from #58 covers it from here on. Plain-text filenames are invisible to that check, which is why this survived.

Verification

• URLs were not retyped. Every link in the new French and Japanese text was substituted from readme.md through a token placeholder ({U1}…{U6}), so translation cannot have corrupted a URL. Verified after the fact: all 28 URLs in readme_fr.md appear verbatim in readme.md. (readme_jp.md has 16 that do not — those are its own pre-existing Amazon links in the purchased-parts table, untouched here; git diff adds none of them.)
• Row pairing for the French relink was guarded, not assumed. Rows are matched on the numeric tokens in the spec cell, because the prose is translated — "boss type, 16 teeth, 6 mm bore" vs "type moyeu, 16 dents, alésage 6 mm" share [1, 16, 6]. The script aborts on any mismatch rather than writing a wrong link; it did abort on the first attempt, which is how the guard got built.
• Structure now matches: all five files have 12 tables with identical row counts, except the Chinese purchased-parts table's 28th row — which is the open question below, not an oversight.
• tools/check_readme_sync.py still passes, and a case-sensitive relative-link audit reports 0 failures, including the new D435_Gemini2_Mount.step links.

What I deliberately did not do

Two BOM contradictions need someone who knows the hardware, and I am not going to guess at either:

• Dowel-pin length — English, Spanish, French and Japanese say M4*8mm; Chinese says M4*7mm. The English row's own Amazon link is a search whose keywords are Dowel+pin+M4*7, which points at M4×7 being right, but that is inference. (Also: the M4×8 and M4×12 rows link to the same Amazon product, so at most one of those links is correct.)
• CAN power-distribution board — only the Chinese BOM lists it (¥9). Either four BOMs are missing a component or one has an obsolete row.

Both are described in #64. Tell me which way they go and I will send the follow-up.

Note on the translations

I wrote the French and Japanese text. The tables are largely part names, dimensions and prices, and the step captions are short imperatives, so the risk is low — but a native reader's glance over the assembly captions would be worth having before merge.

🤖 Generated with Claude Code