DecimalFloat was never deployed at its currently pinned address: 5 red testProdDeployment* tests need an on-chain deploy, not a code change

[thedavidmeister]

LibDecimalFloatDeployProdTest has five red tests on main and every open
branch, and no amount of in-repo work can fix them:

[FAIL: arbitrum: DecimalFloat not deployed]     testProdDeploymentArbitrum()
[FAIL: base: DecimalFloat not deployed]         testProdDeploymentBase()
[FAIL: base_sepolia: DecimalFloat not deployed] testProdDeploymentBaseSepolia()
[FAIL: flare: DecimalFloat not deployed]        testProdDeploymentFlare()
[FAIL: polygon: DecimalFloat not deployed]      testProdDeploymentPolygon()

The pinned address has no code at it

• eth_getCode on the current pin 0x799632d282178e770C7465cad54aDA1021A913D6
  returns 0x on arbitrum, base and flare.
• eth_getCode on the previous 0.1.1 pin
  0xBee0eEcB2f2b3aE6E5Aa9E2C8C0EE9b5eDe9C926 returns 39261 hex chars of code
  on all three. That one is real; the current one was never deployed.

The constants are not wrong — the deployment is missing

This is the important part, because it rules out the tempting fix.

• A local build of main reproduces keccak256(deployedBytecode) for
  DecimalFloat as 0xdc468883…f55e, which is exactly
  DECIMAL_FLOAT_CONTRACT_HASH. The pins match the source they were derived
  from.
• git diff sol-v0.1.7 origin/main -- src/ is empty, so 0.1.7's codehash is
  that same value.

So the source, the codehash and the derived address all agree with each other.
What is missing is a broadcast.

How it got here

Commit 4d98fce (2026-06-17, "after formatter fix") moved the constants to a
newly derived address, as it should have. The Manual sol artifacts dispatch
on 2026-06-29 then tried to deploy it and reverted at gas estimation
(execution reverted, data: 0x), so the address the repo now points at has
never had code at it.

What must NOT be done

• Reverting the pins to 0xBee0eE…C926 would turn CI green while silently
  repointing every downstream consumer at a different mainnet contract with
  different bytecode. That is a production change disguised as a CI fix.
• Weakening or deleting LibDecimalFloatDeployProdTest is the same act in
  a different place. The test is correct: it is reporting a true fact.

What is needed

A deployment of DecimalFloat across arbitrum, base, base_sepolia, flare and
polygon, at the current bytecode, via the usual dispatch:

gh workflow run manual-sol-artifacts.yaml --ref main -f suite=decimal-float

The 2026-06-29 gas-estimation revert should be expected to recur and needs
diagnosing as part of this — deploying is not simply a matter of re-running
the dispatch.

Until then rainix-sol / test stays red on these five regardless of every
other fix, so main cannot go fully green without a human deploying.

Filed while working through the other causes of red CI (#265, #266, #267,
#268), none of which can touch this one.