[PERF] evaluation: args definitions at compile time

[LucasLefevre]

Description:

A number of checks and pre/post processing is done for each executed
function.
Some of it can be known at compile time since it only depends on the
function and the number of arguments.

With this commit, we pre-compute a number of things only once at compile
time to minimize what is done when executing the function.

Bunch report

cells imported in
09ccc24: Mean: 2841.65 ms, StdErr: 3.91 ms
c0ed960: Mean: 2758.60 ms, StdErr: 6.32 ms → 🟢 (vs prev: -3%, Δ=-83.05 ms, combined StdErr=7.43 ms, |Δ|/SE=-11.2x; n=20)

evaluate all cells
09ccc24: Mean: 2678.65 ms, StdErr: 8.38 ms
c0ed960: Mean: 2458.64 ms, StdErr: 11.72 ms → 🟢 (vs prev: -8%, Δ=-220.02 ms, combined StdErr=14.41 ms, |Δ|/SE=-15.3x; n=20)

Model created in
09ccc24: Mean: 6470.76 ms, StdErr: 9.44 ms
c0ed960: Mean: 6158.89 ms, StdErr: 14.90 ms → 🟢 (vs prev: -5%, Δ=-311.87 ms, combined StdErr=17.64 ms, |Δ|/SE=-17.7x; n=20)

Large formula dataset

cells imported in
09ccc24: Mean: 2703.84 ms, StdErr: 14.52 ms
c0ed960: Mean: 2704.29 ms, StdErr: 14.18 ms → ⚫ (vs prev: +0%, Δ=+0.45 ms, combined StdErr=20.30 ms, |Δ|/SE=0.0x; n=20)

evaluate all cells
09ccc24: Mean: 400.51 ms, StdErr: 2.94 ms
c0ed960: Mean: 375.49 ms, StdErr: 2.43 ms → 🟢 (vs prev: -6%, Δ=-25.02 ms, combined StdErr=3.81 ms, |Δ|/SE=-6.6x; n=20)

Model created in
09ccc24: Mean: 3441.85 ms, StdErr: 14.57 ms
c0ed960: Mean: 3419.21 ms, StdErr: 14.61 ms → ⚫ (vs prev: -1%, Δ=-22.64 ms, combined StdErr=20.63 ms, |Δ|/SE=-1.1x; n=20)

Other production spreadsheet (heavily vectorized)

⚫ no measureable change anywhere

Yet another production spreadsheet (with some vectorization)

cells imported in
09ccc24: Mean: 491.04 ms, StdErr: 6.53 ms
c0ed960: Mean: 493.73 ms, StdErr: 4.17 ms → ⚫ (vs prev: +1%, Δ=+2.69 ms, combined StdErr=7.75 ms, |Δ|/SE=0.3x; n=20)

evaluate all cells
09ccc24: Mean: 660.53 ms, StdErr: 3.85 ms
c0ed960: Mean: 633.88 ms, StdErr: 3.46 ms → 🟢 (vs prev: -4%, Δ=-26.65 ms, combined StdErr=5.18 ms, |Δ|/SE=-5.1x; n=20)

Model created in
09ccc24: Mean: 1423.05 ms, StdErr: 9.20 ms
c0ed960: Mean: 1396.71 ms, StdErr: 7.02 ms → 🟢 (vs prev: -2%, Δ=-26.33 ms, combined StdErr=11.57 ms, |Δ|/SE=-2.3x; n=20)

Legend:
⚫: no measurable change |Δ|/SE < 2
🔴: slower Δ > 0 and |Δ|/SE >= 2
🟢: faster Δ < 0 and |Δ|/SE >= 2

Task: TASK_ID

review checklist

• feature is organized in plugin, or UI components
• support of duplicate sheet (deep copy)
• in model/core: ranges are Range object, and can be adapted (adaptRanges)
• in model/UI: ranges are strings (to show the user)
• undo-able commands (uses this.history.update)
• multiuser-able commands (has inverse commands and transformations where needed)
• new/updated/removed commands are documented
• exportable in excel
• translations (_t("qmsdf %s", abc))
• unit tested
• clean commented code
• track breaking changes
• doc is rebuild (npm run doc)
• status is correct in Odoo

[robodoo]

[laa-odoo]

Review taken into account,
thanks @LucasLefevre