Update dtk-template

[LagoLunatic]

• Progress upload to frogress has been removed.
• sjiswrap updated to support .inc files, so d_event_debug can be fixed now.
• objdiff 3 changed its algorithm so the bot will report a lot of fake diffs for this PR.

Note that I also enabled functionRelocDiffs=data_value in the progress report config, which allows for more types of regressions to be caught by diffing progress reports which would currently slip through in some cases that don't affect codegen, such as calling the wrong function symbol or using the wrong float literal. This is optional, can be disabled by commenting that line out if it causes any unwanted behavior.

[decomp-dev]

Report for GZ2E01 (3c2a99e - d7b5e28)

Total code: 11487996 bytes (-368 bytes)
Total functions: 48107 (-73)
📉 Matched code: 61.14% (-0.00%, -512 bytes)
📈 Linked code: 26.89% (+0.11%, +11976 bytes)

📉 framework/m_Do/m_Do_printf | OSSwitchFiberEx(unsigned long, unsigned long, unsigned long, unsigned long, unsigned long, unsigned long) -1 bytes -> 8.33%
📈 framework/m_Do/m_Do_machine | mDoMch_Create() +34 bytes -> 99.78%
📈 framework/m_Do/m_Do_ext | mDoExt_McaMorf::calc() +248 bytes -> 98.32%
📈 framework/m_Do/m_Do_ext | mDoExt_McaMorfSO::calc() +248 bytes -> 98.32%
📈 framework/m_Do/m_Do_ext | mDoExt_McaMorf2::calc() +178 bytes -> 99.37%
📈 framework/m_Do/m_Do_ext | mDoExt_3Dline_c::init(unsigned short, int, int) +26 bytes -> 99.73%
📈 framework/m_Do/m_Do_ext | mDoExt_3DlineMat0_c::draw() +22 bytes -> 99.59%
📉 framework/m_Do/m_Do_ext | mDoExt_3DlineMat0_c::update(int, float, _GXColor&, unsigned short, dKy_tevstr_c*) -1 bytes -> 0.27%
📈 framework/m_Do/m_Do_ext | mDoExt_3DlineMat0_c::update(int, _GXColor&, dKy_tevstr_c*) +371 bytes -> 94.70%
📉 framework/m_Do/m_Do_ext | mDoExt_3DlineMat1_c::init(unsigned short, unsigned short, ResTIMG*, int) 0 bytes -> 0.98%
📈 framework/m_Do/m_Do_ext | mDoExt_3DlineMat1_c::draw() +31 bytes -> 98.85%
📉 framework/m_Do/m_Do_ext | mDoExt_3DlineMat1_c::update(int, _GXColor&, dKy_tevstr_c*) 0 bytes -> 0.21%
📉 framework/d/d_com_static | __arraydtor$4479 -1 bytes -> 99.14%
📉 framework/d/d_com_static | __arraydtor$3839 -1 bytes -> 99.86%
📈 framework/d/d_demo | dDemo_actor_c::getDemoIDData(int*, int*, int*, unsigned short*, unsigned char*) +20 bytes -> 99.94%
📈 framework/d/d_demo | dDemo_setDemoData(fopAc_ac_c*, unsigned char, mDoExt_McaMorf*, const char*, int, unsigned short*, unsigned long, signed char) +27 bytes -> 99.15%
📈 framework/d/d_resorce | addWarpMaterial(J3DModelData*) +5 bytes -> 98.87%
📈 framework/d/d_map_path_dmap | renderingPlusDoorAndCursor_c::drawTreasure() +4 bytes -> 99.53%
📈 framework/d/d_map_path_dmap | renderingPlusDoorAndCursor_c::drawTreasureAfterPlayer() +4 bytes -> 99.53%
📈 framework/d/d_particle | dPa_gen_b_light8PcallBack::draw(JPABaseEmitter*, JPABaseParticle*) +12 bytes -> 99.96%
📈 framework/d/d_kankyo_rain | dKyr_rain_move() +209 bytes -> 99.81%
📈 framework/d/d_kankyo_rain | dKyr_snow_move() +8 bytes -> 99.98%
📈 framework/d/d_kankyo_rain | vrkumo_move() +17 bytes -> 99.46%
📈 framework/d/d_kankyo_rain | dKyr_drawLenzflare(float(*)[4], cXyz*, _GXColor&, unsigned char**) +487 bytes -> 98.59%
📈 framework/d/d_kankyo_rain | dKyr_drawSnow(float(*)[4], unsigned char**) +8 bytes -> 99.84%
📈 framework/d/d_kankyo_rain | drawVrkumo(float(*)[4], _GXColor&, unsigned char**) +440 bytes -> 99.43%
📈 framework/d/d_kankyo_rain | dKyr_odour_move() +14 bytes -> 99.90%
📈 framework/d/d_kankyo_rain | dKyr_odour_draw(float(*)[4], unsigned char**) +68 bytes -> 99.83%
📈 framework/d/d_bg_w_kcol | dBgWKCol::LineCheck(cBgS_LinChk*) +317 bytes -> 98.04%
📈 framework/d/d_bg_w_kcol | dBgWKCol::GroundCross(cBgS_GndChk*) +26 bytes -> 99.37%
...and 5905 more items

[TakaRikka]

are there any important regressions to note? matched code dropping 23084 bytes seems like a lot

[LagoLunatic]

are there any important regressions to note? matched code dropping 23084 bytes seems like a lot

I fixed most of the 100% regressions, it was pretty much symbols and splits being wrong because they're missing from TP's maps. Also linked the TUs now that the splits are fixed.

I don't understand what the last three 100% regressions are (those -512 bytes):

$ ninja changes | grep 100\\.00%
                                                                        framework/d/actor/d_a_npc_cd | matched_data | 100.00% ->  99.97%
                                                                        framework/dolphin/os/__start | fuzzy_match  | 100.00% ->  99.84%
                                                                        framework/dolphin/os/__start | matched_code | 100.00% ->  36.92%
                                                                                             __start | fuzzy_match  | 100.00% ->  99.66%
                                                                                         __init_data | fuzzy_match  | 100.00% ->  99.90%

They show seem to 100% in objdiff GUI, so I'm not sure how objdiff-cli is counting them.